REACTOR PHYSICS

COURSE INTRODUCTION
This Training Manual assumes prior knowledge of Nuclear Theory. It extends this information
into a discussion of Reactor Physics, particularly as it relates to CANDU reactors.

The course begins with the general principles of reactor configuration required to maintain a self-
sustaining chain reaction. It continues with reactor dynamics (in both the critical and subcritical
core), reactivity feedback effects (temperature effects, fission product poisoning, and fuel
burnup), and ends with operational considerations (at low and high power).
The material covers four main areas, subdivided into eight major sections
as follows:
• The Critical Reactor at Steady Power Output (Section 1)
• The Dynamic Reactor (Sections 2 and 3)
• Reactivity Feedback Effects (Sections 4, 5, and 6)
• Reactor Operations (Sections 7 and 8)
CNSC i
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

TABLE OF CONTENTS
Objectives 1
The Critical Reactor at Steady Power Output 9
1.0 INTRODUCTION 9
1.1 Fission 9
1.2 Harnessing Fission 15
1.3 Movement of Neutrons Through the CANDU lattice Lattice 18
1.4 The Finite Reactor 26
Response of The Critical Reactor to a Reactivity Change 31
2.0 INTRODUCTION 31
2.1 Exponential Power Rise 31
2.2 Corrections to Exponential Reactor Response 33
2.3 The Effect of Delayed Neutrons 36
2.4 Prompt Criticality 40
2.5 Power Rundown: The Prompt Drop 42
Responsiveness of The Subcritical Reactor 45
3.0 INTRODUCTION 45
3.1 Neutron Flux in a “Shut Down” Reactor 45
3.2 Dynamics in the Subcritical Core 50
3.3 Examples 55
Effects of Temperature and Voiding on Core Reactivity 61
4.0 INTRODUCTION 61
4.1 Feedback—An Introduction to Temperature Effects e 61
4.2 The Physical Basis for Temperature Coefficients 63
4.3 Temperature Coefficients of Reactivity 72
4.4 Reactivity Variation with Temperature 77
4.5 Void Reactivity 79
Effects of Fission Products on Core Reactivity 85
5.0 INTRODUCTION 85
5.1 Xenon and Iodine Buildup 85
5.2 Transient Xenon Behaviour 93
CNSC ii
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

5.3 Xenon Oscillations 103
5.4 Samarium-149 106
Effects of Fuel Irradiation and On-Power Fuelling on Core Reactivity 111
6.0 INTRODUCTION 111
6.1 On-Power Fuelling 111
6.2 Fuel Burnup—General 112
6.3 Transient Reactivity Changes 115
6.4 Long-Term Reactivity Effects 115
Reactor Operations at Low Power 125
7.0 INTRODUCTION 125
7.1 Thermal Power, Neuron Power, AND Fission Power 125
7.2 Reactor Power Rundown 126
7.3 The Shutdown State 133
7.4 Approach to Critical 135
7.5 Low Power Operation Following Startup 142
Reactor Operations at High Power 145
8.0 INTRODUCTION 145
8.1 Flux Flattening 145
8.2 Flux Shape Details 150
8.3 High Power Protection 161


CNSC 1
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

OBJECTIVES
At the end of training, the participants will be able to:
The Critical Reactor at Steady Power Output
List the reaction products of the fission process and for each describe its
importance for CANDU operation.
1.0 Describe the characteristics of fission products with respect to yield,
stability, radiation hazard, delayed neutron production, and capability
to absorb neutrons.
2.0 Identify the different energy contributions that make up the
approximately 200 MeV per fission deposited in the reactor.
3.0 Define the terms:
- unit cell multiplication factor,
- effective neutron multiplication factor.
4.0 Describe the neutron life cycle in terms of the following neutron
processes:
- fast fission of U-238,
- resonance absorption in U-238,
- absorption in non-fuel core materials,
- thermal fission following absorption in fuel,
- neutron leakage.
5.0 Define reactivity and give its units of measurement.
6.0 Define the terms:
- reactivity worth,
- excess core reactivity,
- nominal core,
- control reactivity.
Response of the Critical Reactor to a Reactivity Change
7.0 Define log rate and reactor period and state their relationship
8.0 Describe the response of a CANDU reactor at low power to a small
step insertion of positive reactivity
9.0 Explain the effect of delayed neutrons on reactor control.
10.0 Define the following and state how they arise:
- prompt jump,
- prompt drop.
CNSC 2
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

11.0 Describe the change in reactor power following a positive
reactivity insertion large enough to cause a prompt jump and state why
power first rises rapidly and then increases more slowly.
12.0 Describe the change in reactor power following a large negative
reactivity insertion and state why the power first drops rapidly and
then decreases more slowly.
13.0 Define prompt criticality, explain how it arises, and state the
approximate reactivity insertion required to cause prompt criticality.
Responsiveness of the Subcritical Reactor
14.0 Define the subcritical multiplication factor.
15.0 Explain how subcritical multiplication of a neutron source in a
subcritical core causes:
- an observable, steady power level that is larger than the source,
- a change in power level after a reactivity change that leaves the
core subcritical.
• Describe the rate of response to a reactivity change in a subcritical
core.
• Indicate how and why the dynamic response changes in going from
a deeply subcritical reactor to an almost critical reactor.
Effects of Temperature and Voiding on Core Reactivity
• Define temperature coefficients of reactivity for the:
- fuel,
- moderator, and
- heat transport coolant.
• State the typical operating temperatures for the fuel, moderator,
and coolant and the approximate range of temperatures encountered in
going from cold shutdown to full power.
• Explain how thermal expansion of the moderator affects:
- neutron path lengths in the moderator,
- leakage from the core.
• Explain how increased molecular speeds caused by heating affect
the:
- resonance Absorption in U-238 (Doppler broadening and self
shielding),
- thermal neutron spectrum, and
- thermal neutron path lengths.
CNSC 3
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


• Describe how changes in the thermal neutron temperature
affect absorption in:
- U-235,
- Pu-239.
16.0 Explain the effect on reactivity caused by a change in temperature
of the:
- fuel,
- moderator,
- coolant.
• Given typical values of moderator, HT coolant and fuel reactivity
coefficients, calculate the reactivity change that would occur for
typical unit operations, including:
- HT system warm up from cold to zero power hot,
- increase in unit power from zero to full power, and
- decrease in unit power from full power to zero.
• Define Power Coefficient
• Compare the sizes of the reactivity changes due to the moderator,
coolant, and fuel for a given change in reactor power with reference to:
- temperature coefficient,
- temperature change,
- time for the effect to show up.
Describe the effect of a typical CANDU power coefficient on:
- normal regulation,
- a power transient following an upset.
• Define the term “void reactivity”.
• Explain how voiding of the coolant simultaneously increases fast
fission and decreases resonance capture.
• Describe how the thermal neutron temperature changes on coolant
voiding and state the effects on fission rate.
• Explain how void reactivity leads to a lower limit on coolant
isotopic.
• Explain the upper limit on heat transport isotopic.
Effects of Fission Products on Core Reactivity
• Define the term, fission product poison.
• State the characteristics of xenon-135 that make it a significant
fission product for reactor operations.
CNSC 4
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


• Describe how nuclear processes in the fuel:
- produce I-135 and Xe-135,
- remove I-135 and Xe-135.
• Define xenon load and iodine load.
• Explain the given curve shapes for:
- Iodine load vs. time,
- Xenon load vs. time.
for start up after a long shutdown.
17.0 Describe xenon buildup following a trip.
18.0 State the operational problem xenon causes when the reactor trips.
• Explain the following terms:
- poison out,
- poison prevent operation,
- poison override capability,
- poison override time,
- decision and action time.
19.0 Explain the following features of the reactivity change following a
trip from full power:
- initial rate of xenon buildup,
- the peak,
- eventual decrease in xenon concentration.
20.0 State the approximate time for xenon to peak and the approximate
reactivity worth of xenon at the peak for a reactor trip from full power
with equilibrium fuel.
21.0 Compare the sizes of the transient peaks following a reactor trip
from equilibrium conditions for a trip from full power and a trip from
a lower power level.
• Explain the power transient that follows a change in reactor power
in the high power range.
• Describe how the reactivity transient is counteracted after a return
to power:
- before a poison out,
- following a poison out.
• Define the terms:
- xenon oscillation,
- flux tilt.
• State why large oscillations are unacceptable
• Describe how oscillations are controlled in CANDU reactors.
CNSC 5
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

• Explain how a small, local, reactivity change can cause a large flux
tilt in a reactor operating at high power without adequate spatial
control.
• Describe how a flux tilt changes with time if left to itself.
• Explain why an uncontrolled oscillation may continue indefinitely.
• State the characteristics of the samarium-149 that make it a fission
product poison.
• Describe how nuclear processes in the fuel:
- produce Pm-149 and Sm-149,
- remove Pm-149 and Sm-149.
• Compare the operational effects of samarium with those of xenon
with respect to:
- initial build up,
- transient following a trip or shutdown,
- return to equilibrium following a restart,
- transients on power changes.
Effects of Fuel Irradiation and On-Power Fuelling on Core Reactivity
• List the chief characteristics associated with using on-power
fuelling for maintaining core reactivity.
• Define the terms:
- fresh fuel,
- fuel burnup, and
- equilibrium-fuelled reactor.
• State and explain the fuel burnup units
22.0 Describe the transient change in Pu-239 following a shutdown
from high power, and following return to high power after a shutdown.
• Define the terms “saturating fission product” and “non-saturating
fission product” and compare their long-term effects on the reactivity
worth of a fuel bundle.
• Describe the changes that occur in the composition of a fuel bundle
as it is exposed to neutron flux in the core.
23.0 Given a graph showing the reactivity change of a fuel bundle with
irradiation, explain the shape of the graph in terms of:
- U-235 burnup and Pu-239 growth,
- buildup of Pu-240 and Pu-241,
- increasing fission products.
CNSC 6
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Reactor Operations at Low Power
24.0 State and explain the reasons for non-linearity between changes in
neutron power and changes in reactor thermal power.
Explain the size and duration of the prompt drop given a curve of neutron
flux decrease following a trip.
Describe how prompt neutrons, delayed neutrons, photoneutrons, and
spontaneous fission neutrons affect the shape of the rundown curve.
25.0 Explain the following differences between the thermal power and
neutron power rundown curves following a trip:
- initial rate of drop,
- duration,
- cooling requirements.
26.0 State the approximate value of decay heat at full power, and at 3
minutes and 60 minutes after a trip from full power.
• Identify the reactivity changes that occur in a reactor after it is
shutdown from extended operation at power and, for each change give
the:
- sign,
- approximate size,
- time scale.
27.0 Describe the variation in neutron flux while the reactor is in a
shutdown condition for a long period of time.
28.0 List the parameters specifically related to criticality that are
monitored and controlled during the approach to critical.
29.0 List reactivity control mechanisms required during the approach to
criticality.
30.0 Explain (for step reactivity increases) the observed changes in:
- stable count rate,
- flux detector time response,
as the core is taken from deeply subcritical to almost critical.
• Predict the power level following a specified change of poison
concentration, given the subcritical multiplication formula, the
measured power, and the poison concentration relative to criticality.
31.0 Explain why start-up procedures require monitoring of neutron
flux during start-up and do not depend solely on criticality predictions.
32.0 List parameters that should be monitored and controlled on
reaching criticality.
CNSC 7
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


33.0 Explain in general terms how a reactor, critical at low power, could
become subcritical:
- if held in the low power “critical” state for a long time,
- following power manoeuvres at low power.
34.0 Describe the power response of a slightly supercritical reactor at
low power.
35.0 Explain the change in power response as power approaches the last
decade of reactor power.
• Describe how reactor power is increased to rated power from
hot shutdown.
36.0 List the reactivity effects that occur as power rises and state how
they are expected to change as power increases.
Reactor Operations at High Power
37.0 Explain how flux-flattening permits increased reactor power output
without an unsafe increase in peak bundle or channel power.
38.0 Describe and explain how flux flattening is achieved with respect
to:
- the overall flux shape,
- relative zone to zone flux shape,
- control of local flux peaks.
39.0 Explain how a change of reactivity device configuration can cause
flux peaks in the core and how this can affect bundle and channel
powers.
40.0 Define the terms:
- reactivity device worth,
- differential reactivity device worth.
41.0 Explain the variation in rate of reactivity insertion given a curve of
rod worth vs. position.
42.0 Describe the response of the liquid zones while the adjusters are
driving from full in to full out.
43.0 Describe the effect on flux shape of replacing high burnup fuel
with fresh fuel while operating at power.
44.0 Explain how shutdown fuelling could lead to an unacceptable
spatial flux distribution on restart.
45.0 Describe what the “Reference Flux Shape” means and explain why
the flux shape in an operating CANDU differs from the reference
shape.
CNSC 8
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


46.0 Define the following terms as they apply on an operating CANDU
reactor:
- fuelling ripple,
- channel power peaking factor (CPPF).
47.0 State why bulk power is limited when the reactor is operating with
adjusters out of core.
48.0 State how the reactor is protected from excessive high power when
operating in an unanalyzed flux shape.
49.0 State the approximate power levels above which reactor operation
is affected by:
- transient xenon,
- xenon oscillations.

Notes:
CNSC 9
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


THE CRITICAL REACTOR AT STEADY POWER OUTPUT
1.0 INTRODUCTION
The defining characteristic of a nuclear reactor is the fission chain reaction. When
an atom such as uranium fissions (splits) it releases energy, and the amount it
releases, per reaction, is huge compared to a chemical reaction. This translates
into a power source that has relatively low fuel cost. A crucial feature of the
fission reaction is that it generates neutrons that can cause fission. When at least
one fission neutron per fission survives to cause fission, the chain reaction sustains
itself without any help from an outside source of neutrons; it is a self-sustaining
chain reaction.
A nuclear reactor is a mixture of nuclear fuel, moderator, coolant, and structural
material arranged in a suitable configuration of sufficient size, to support a
controllable, self-sustaining chain reaction that releases useful energy. The fission
reaction also produces radiation and radioactive by-products that present an
immediate hazard. This hazard defines many characteristics of reactor design and
operation.
Natural uranium (UO
2
) fuel and heavy water (D
2
O) moderator characterize
CANDU reactors. (The name derives from the words CANada, Deuterium, and
Uranium). Requirements related to these design choices resulted in a pressure-
tube reactor with on power fuelling: a reactor that is neutron efficient (wastes very
few neutrons), with a well-thermalized neutron spectrum. This section reviews the
fission process and examines the balance between neutron production and loss in
the reactor core that allow for a sustained rate of fission. It then describes how the
CANDU characteristics affect neutrons as they travel through the CANDU lattice;
effects that help explain features of day-to-day operation described in later
sections.
To discuss nuclear processes that go on in a reactor, the trainee must be acquainted
with some basic nuclear theory about the structure of the atomic nucleus and the
principles governing the release and deposition of energy in nuclear processes
such as radioactivity and fission. These notes assume that previous courses have
presented the required background information, including the description of atoms,
nuclei, atomic particles, mass-energy equivalence, and the interaction of radiation
with matter.
1.1 FISSION
Nuclear power production relies on the fact that some nuclei will fission, releasing
energy (about 200 MeV per fission) because, in splitting into more tightly bound
fragments, some of the nuclear mass becomes energy.
Notes:
CNSC 10
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Each watt of power requires about 3.1 × 10
10
fissions/s. For a CANDU
reactor operating at 2250 MW, the required fission rate is a steady:
2250 × 10
6
× 3.1 × 10
10
= 7 × 10
19
fissions/s
The fission process provides the neutrons it needs to keep itself going. Each
fission releases, on average, about 2.5 neutrons. A steady fission rate requires that
neutron losses by capture (radiative capture in fuel or other materials) or by
leakage (escape from the reactor) not exceed about 1.5 neutrons per fission.
Reaction Products of the Fission Reaction
The particular mode of fission illustrated in Figure 1.1 is one of many possible
ways in which the nucleus may divide. The fission fragments Xe-140 and Sr-96
are two of about 300 nuclides that fission produces.
Initially the fission fragments are highly excited and get rid of excitation energy by
emitting neutrons and gamma rays. Most of the fission neutrons are emitted
almost immediately

(~10
-17
s) after the fission takes place; these are known as
prompt fission neutrons, while the gamma rays are known as prompt gammas.
Following de-excitation, a nucleus formed by fission does not have sufficient
excitation energy to throw off any more neutrons. It usually decays by beta
emission; Figure 1.1 shows an example decay chain. In a few cases, the daughter
nucleus created by beta decay has sufficient energy to emit a neutron. This
process results in approximately ½% or so of the fission neutrons showing up long
after the fission event. These are the delayed fission neutrons.


n + U
235
92
139
54
Xe Cs Ba La
139
55
139
56
139
57
γ γ γ
β− β− β−
n + n + γ (from precursors of Xe-139 and/or Sr-95)
95
38
Sr Y Zr Nb Mo
95
39
95
40
95
41
γ γ γ
β− β− β−
γ
β−
95
42

Figure 1.1
Fission Product Decay Chains
The standard notation for the particular fission just described is as follows:
γ + + + → → + n 2 Sr Xe U U n
95
38
139
54
* 236
92
235
92

Notes:
CNSC 11
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

After U-235 absorbs the neutron, the compound nucleus (
* 236
92
U ) survives
for about 10
-14
s, and then fission occurs about 85% of the time. (Radiative
capture occurs about 15% of the time). The equation displays the immediate break
up, with prompt neutrons and prompt gammas. Leaving out the compound
nucleus simplifies the notation:
γ + + + → + n 2 Sr Xe U n
95
38
139
54
235
92

The energy release per fission, averaged over all fission reactions in a reactor core,
is about 200 MeV (corresponding to a mass loss of nearly 0.2 u). This is,
coincidently, approximately the energy per fission deposited in the reactor core.
Individual fission reactions usually do not deviate from 200 MeV by more than a
couple MeV, making it a convenient benchmark for later calculations. About
180 MeV of the energy release occurs at the moment of fission, the moment
displayed in the notation just given. The beta emissions that follow release the
remaining energy. Table 1.1 shows the distribution of energy released over the
reaction products, and shows the energy deposition in the core.
Notes:
CNSC 12
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003



Energy
Released
Energy
Deposited in Core
Energy Source
Energy
(MeV)
Energy
(MeV)
Energy
(%)
Kinetic energy of
lighter fission fragment 100 MeV 100 MeV 50%
Kinetic energy of
heavier fission fragment 67 MeV 67 MeV 33.5%
Kinetic energy of
fission neutrons 5 MeV 5 MeV 2.5%
Energy of prompt gamma rays 6 MeV 6 MeV 3%
Radiative capture of
excess fission neutrons — 8 MeV 4%
Beta particle energy gradually
released from fission products 8 MeV 4%
Gamma ray energy gradually
released from fission products 6 MeV 3%
Neutrinos
22 MeV
(shared by
β
-
particles,
gamma
rays and
neutrinos)
— —
Total 200 MeV 200 MeV 100%
Table 1.1
Approximate Energy Distribution Resulting from a Typical U-235 Fission
The number of neutrons emitted following fission is variable, but the most
probable yields are two or three neutrons per fission. The average number of
neutrons emitted per thermal fission is an important quantity in reactor physics.
The Greek letter ν (pronounced “new”) usually denotes it. Table 1.2 summarizes
the values of ν for the fissile isotopes. Fast fission produces marginally more
fission neutrons than thermal fission. These notes assume that the average number
of neutrons per thermal fission in an equilibrium fuelled reactor is ν = 2.5.
Notes:
CNSC 13
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003



Fissile Nucleus ν
U-235 2.43
Pu-239 2.89
Pu-241 2.93
Equilibrium Fuel ~ 2.5
Table 1.2
Value of ν for thermal fissions
Fission Fragments and Fission Products
The name “fission products” applies to all members of the decay chains—
including the original fission fragments. The term fragment describes the initial
flying pieces that de-excite before they come to rest as fission products. The
majority of fission products have half-lives that range from fractions of a second to
about 30 years.
The kinetic energy of the fission fragments accounts for most of the energy
released in the fission reaction. The energetic fragments slow and stop in about
10
-13
s, (after travelling about 10
-3
mm), heating the fuel by passing their energy to
the surrounding fuel atoms.
Each fission product consists of a substantial piece of the original nucleus. They
are likely to have mass numbers between 70 and 160, with mass numbers near 95
and 140 the most probable. Symmetrical fission (equal fragments) is rare.
Fission products are a potential radiological hazard. They also absorb neutrons
that might otherwise cause fission, limiting the long-term energy output from the
fuel. They have one important useful effect; reactor control would not be possible
without the delayed neutrons that result from fission product decay. The next few
paragraphs discuss these highlighted characteristics one at a time.
The Radiological Hazard from Fission Products
The fission fragments are almost invariably unstable (radioactive). The reason for
this is that the neutron/proton ratio of the fragments is about the same as that of the
fissioned nucleus, and this is too high for stability at medium mass numbers.
Standard beta decay notation records the subsequent beta decays.

Notes:
CNSC 14
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

For example, following one chain of Figure 1.1:
|
.
|

\
|
= γ + β + →
|
.
|

\
|
= γ + β + →
|
.
|

\
|
= γ + β + →



h 4 . 1 t La Ba
m 3 . 9 t Ba Cs
s 40 t Cs Xe
2
1
2
1
2
1
139
57
139
56
139
56
139
55
139
55
139
54

The ceramic fuel material (UO
2
) and fuel sheathing must encase the fission
products to prevent them entering the heat transport system and leaving the reactor
core. (There are additional barriers in place to limit release to the public if the
piping fails.) As long as the fission products remain in the fuel and the fuel
remains adequately shielded, there is no biological risk. Fission products in the
heat transport system (or elsewhere outside the core) are a radiation hazard that
would prevent equipment access even with the reactor shut down.
Heavy shielding is required around the reactor for protection against the prompt
radiation (neutrons and gamma rays) during operation. It also limits exposure to
gamma radiation that continues to be emitted by the fission products after
shutdown.
Fuel must be replaced remotely; special precautions must be taken in handling,
and storing spent fuel.
Neutron Absorption by Fission Products
Some of the fission products have high neutron absorption cross-sections and
thereby poison the reactor. A relatively high percentage of fissions produce the
two most important poisons, Xe-135 and Sm-149, and these fission products
capture a significant number of neutrons. Section 5 examines the effects of fission
product poisons.
Delayed and Photoneutrons from Fission Product Decay
Fission product decay sometimes produces delayed neutron precursors, which in
decaying produce a small fraction of the neutrons in the reactor core. Energetic
gamma rays from decaying fission products may occasionally produce photo-
neutrons. Fission neutrons make possible a self-sustaining chain reaction, in
which the fission neutrons from one generation of fissions cause the next
generation of fissions. Delayed neutrons make reactor control possible (See
Section 2). Photoneutrons affect low power operation (See Section 3).
The Chart of Nuclides identifies nearly one hundred delayed neutron precursors.
For thermal fission of U-235, the delayed neutrons are only 0.70% of all fission
Notes:
CNSC 15
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

neutrons. This fraction, represented by the symbol β, is called the delayed
neutron fraction. For Pu-239, the delayed neutron fraction is even smaller, at β =
0.23%.
For equilibrium CANDU fuel, the effective average delayed neutron fraction is
close to β ~ 0.50%. Equilibrium fuel accounts for fissions of all fissile isotopes in
the core, as well as fast fission of U-238. The weighted average depends on the
particular mix of isotopes in the reactor, a mix that changes with fuel burnup.
A set of six artificial delayed neutron precursors accurately models the behaviour
of all the delayed neutrons. These notes make the simplifying assumption of a
single group of delayed neutrons with β = 0.5% for equilibrium fuel, and an
average delay time t
m
= 12.5 s (equivalent to t
½
= 8.7 s and λ = 0.08 s
-1
).
1.2 HARNESSING FISSION
A single fission reaction releases about 40 million times as much energy as the
release of chemical energy that comes from burning a single carbon atom.
However, assembling the reactor materials in a way that allows a continuous,
controllable extraction of energy requires careful design. Neutrons from fission
are essential to maintain a chain reaction and steady power production in the core.
With ν = 2.5, the reactor design must limit neutron losses to 1.5 neutrons per
fission. Absorptions that do not cause fission and neutrons that escape into the
shielding make it difficult to arrange the survival of at least one neutron from each
fission reaction.
For example, consider the possibility of a self-sustaining chain reaction in a large
block of natural uranium. Thermal neutrons absorbed in a block of natural
uranium have about a 55% chance of causing fission. (The ratio of the fission to
absorption microscopic cross-sections for natural uranium is 4.18 b/7.58 b = 0.55).
Since fission produces almost 2.5 new neutrons, this would seem to allow for a
continuous chain reaction. However, fast neutrons produced by fission are less
likely to cause fission than thermal neutrons. (In U-235, the fission cross-section
for thermal neutrons is 580 barns, compared to one or two barns for fast fission).
Now consider what happens as a fast neutron slows down by successive scattering
reactions in the block of uranium. The energy loss per collision is very small,
because the uranium nuclei are so heavy. Before its kinetic energy reduces to
thermal energies, the neutron will reach an energy range where the absorption
cross-section of U-238 is very high and it will be absorbed uselessly. Even
ignoring neutron leakage or absorption in impurities (including absorption in
fission products from those fissions that do occur), the resonance capture loss
makes it impossible to achieve a chain reaction in pure natural uranium.
One solution is to mix the fuel with a moderator that slows the neutrons quickly
through the resonance energy range, without absorbing them once they are
thermalized. For a natural uranium reactor using uranium dioxide fuel, the only
Notes:
CNSC 16
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

possible moderator is heavy water. The choice of the relative volumes of
fuel and moderator and their geometric arrangement determines the neutron
efficiency.
To follow the fate of a neutron as it bounces randomly through a mixture of
reactor materials we need a quantitative comparison of the possible reactions.
Most convenient is a comparison of reaction rates, which depend on nuclear cross-
sections and neutron flux, quantities we will now introduce.
Macroscopic Cross-Section
The rate at which a reaction takes place in bulk material depends on not only the
reaction probability with a particular nucleus (represented by the microscopic
cross-section) but also on how many target nuclei there are. To account for this
we introduce a quantity known as the macroscopic cross-section (Σ—upper case
sigma), which combines the microscopic cross-section (σ) of an individual nucleus
of a given material with the number density, N, (that is, the number of atoms per
cm
3
) of the material in the region considered.
The macroscopic cross-section is given by:
Σ = Nσ
The unit for Σ is cm
2
/cm
3
= cm
-1
(taking N in units of cm
-3
and σ in units of cm
2
).
It is more difficult to visualize the physical significance of the macroscopic cross-
section with its rather odd units than it is for the microscopic cross-section, which
is simply an area (although a somewhat artificial one). One way to think of
macroscopic cross-section is as the total apparent area of all the nuclear targets in
one cm
3
of material. (The blacker the sky is with birds, the less distance a shotgun
pellet is likely to travel into the flock before it makes a hit.)
The inverse of the macroscopic cross section for a given material, which has the
dimensions of a distance, does have an easily visualized meaning. For example,
the quantity 1/Σ
a
equals the average distance that a neutron will travel before being
absorbed by the material, and is known as the absorption mean-free-path, (mfp).
Similarly, the inverse of the macroscopic scattering cross-section, 1/Σ
s
, is equal to
the average distance traveled by a neutron between scattering collisions.
Neutron Flux
The neutron flux (φ) measures the intensity of neutrons passing through a cubic
centimetre of material. It is given by:
φ = nv
where n is the density of neutrons (the number of neutrons per cm
3
) and v is the
speed of the neutrons. The units of flux are:
Notes:
CNSC 17
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


1 2
3
s cm or
s
cm
cm
neutrons
− −
×
Unlike a shotgun, which produces a single “pulse” of pellets that travel together in
more-or-less straight, parallel lines, neutrons travel continuously through the one
cubic centimetre volume in random directions. The chance of a hit depends on
each of the factors n and v. Increasing the number of neutrons in the one cm
3

volume makes it more likely that one of them will hit a target. Increasing neutron
speeds increases the number of visits in the one cm
3
volume each second.
In physical terms, the quantity φ is the total distance traveled in one second by all
the neutrons in the one cm
3
volume, since it is obtained by multiplying the number
of neutrons in that cm
3
by the speed each is travelling. This is equivalent to the
total length of all the neutron tracks laid down in one cm
3
in one second.
Visualize it, if you like, as a time exposure of car taillights taken at night; each
light shows up as a track.
The expression for neutron flux applies to any neutron energy. When applied to
thermal neutrons the product is known as the thermal neutron flux.
Reaction Rate
The reaction rate, R, is the number of reactions per second per cubic centimetre of
material. Consider, for example, a thermal neutron absorption reaction with
microscopic absorption cross section σ
a
representing the apparent absorption size
of each target nucleus. The macroscopic cross section (Σ
a
= Nσ
a
) is the total
apparent flat area of targets that a neutron “sees” as it approaches the one cm
3

volume. The mean free path to absorption is mfp = Σ
-1
. The flux (φ) is the total
neutron track length laid down in one second in one cm
3
, so dividing flux by the
length of track required (on average) for one absorption, we get the total number
of absorptions, that is:
( )
a a a
3
a
R or
mfp
R , so
absorption to path free mean neutron
cm per s per length track total
R
Σ φ
φ
= =
=

With φ in cm
-2
s
-1
and Σ
a
in cm
-1
, R has units cm
-3
s
-1
.
Although we used the thermal neutron absorption reaction for illustration, the
result is perfectly general, for any specified reaction and for neutrons of any
energy:
R = φΣ
Notes:
CNSC 18
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

By far the majority of neutrons in a CANDU core are thermal neutrons,
and rates of reactions induced by these neutrons are directly proportional to both
the thermal neutron flux and the macroscopic cross section for the reaction.
Quoted values of cross-section or flux are always for thermal neutrons unless
otherwise stated.
Tables of microscopic cross-sections usually give values for neutrons travelling at
2200 m/s, that is, with kinetic energy (½mv
2
) of 0.0253 eV. This is the most likely
speed for neutrons thermalized at 20°C. Assuming that the temperatures of
CANDU components are all at 20°C is obviously not physically realistic.
Corrections to the reaction rates for realistic neutron temperatures become
important when we discuss temperature effects in Section 4. As we shall see,
neutron temperature does not strongly affect reaction rates in most reactor
materials. Reaction rates in the fissile isotopes are a notable exception.
1.3 MOVEMENT OF NEUTRONS THROUGH THE CANDU LATTICE
LATTICE
Figure 1.2 shows a neutron following a zigzag path as its direction of motion
changes randomly with each scattering collision. The straight portions of path
between two scattering collisions are usually about 1 to 3 cm in length in most
materials. For reference, the CANDU lattice pitch (centre-to-centre distance of
nearest neighbour channels) is 28.6 cm. The fuel bundle diameter (or pressure
tube inside diameter) is about 10 cm, and the bundle length is close to 0.5 metres.
The fate of a neutron depends on the relative probabilities of the scattering and
absorption reactions for the various nuclei it encounters as it diffuses through the
CANDU lattice. These probabilities depend on the relative abundance of the
different nuclei and the cross-sections for particular reactions.
CANDU fuel begins as uranium dioxide with 99.28% U-238 and 0.72% U-235
(the natural abundance). This new fuel is known as fresh fuel. An equilibrium-
fuelled reactor, in which day-to-day operation includes routine replacement of
“burned up” fuel with fresh fuel, contains both uranium and plutonium. Fuel with
a mix of isotopes comparable to the equilibrium-fuelled core average is called
equilibrium fuel.
In equilibrium, fuel there will be thermal fissions (fissions induced by thermal
neutrons) in U-235, Pu-239, and Pu-241, and fast fissions (fissions induced by fast
neutrons) in U-238. There is a lot of U-238 in the core (more than 100 tonnes in a
large CANDU, compared to less than 1 tonne of the fissile isotopes) but U-238 can
fission only if the neutron kinetic energy is greater than about 1.2 MeV (the
threshold energy for the reaction). There are far fewer fast neutrons than thermal,
so there is relatively little fast fission. (U-238 fast fission contributes less than 3%
of CANDU power production).
Notes:
CNSC 19
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Neutrons of any energy can initiate fission in the fissile isotopes (this is
the meaning of the adjective “fissile”—there is no threshold energy for fission in
U-235, Pu-239, or Pu-241), but the neutron flux and cross-sections are both small
for high-energy neutrons. We can ignore fissions in the reactor core other than
thermal fission of U-235, Pu-239, and Pu-241, and fast fission of U-238.

CANDU moderator is high isotopic heavy water, typically 99.8% or more D
2
O,
with 0.2% or less H
2
O impurity. Similarly, the isotopic of the heavy water coolant
that surrounds the fuel elements in the channels is near 99.8 % isotopic, but
contains slightly more light water impurity than the moderator contains.

Figure 1.2
Typical Path of a Neutron from Birth to Absorption
In Figure 1.2, elastic collisions with nuclei in the moderator slow a neutron born at
the point labelled “fission” until it reaches thermal energy at the point labelled
“neutron thermalized”. In a CANDU reactor, the average distance between these
points is about 25 cm and the time is a few microseconds.
When a fast neutron collides with a nucleus, the fraction of the neutron’s kinetic
energy transferred to the recoiling target nucleus depends on the mass of the target
nucleus and the angle of collision. The average energy transferred on each
collision in the moderator is a fixed fraction of the neutron kinetic energy. In an
elastic collision between a neutron and a heavy water nucleus, the neutron
transfers 40% of its energy, on average, to the target nucleus. Elastic collisions
Notes:
CNSC 20
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

conserve kinetic energy, so the neutron emerges from each “average”
collision with 60% of the energy it started with. To estimate the number of
collisions that will thermalize a fission neutron, assume the fission neutron starts
with a kinetic energy of 2 MeV, loses 60% of its energy on each collision, and is
“thermal” when it has kinetic energy of 0.025 eV. It takes about 36 collisions to
thermalize a fast neutron [0.6
36
× (2 × 10
6
) ~ 0.025].

Following thermalization, the neutron diffuses as a thermal neutron until
something, usually fuel, absorbs it (unless it reaches the edge of the reactor and
leaks into the shielding). The time between thermalization and absorptions
accounts for most of the neutron lifetime, about one millisecond (10
-3
s). The
average distance between birth and absorption in Figure 1.2 is almost 40 cm (close
to the diagonal distance across the CANDU lattice).
The values quoted are “crow-flight” distances—the paths followed by the neutrons
are considerably longer. For example, a thermal neutron moving at 2200 m/s
travels a zigzag distance of 220 cm in 10
-3
s. For a scattering mean free path of
2.9 cm, this implies approximately 75 elastic collisions while diffusing as a
thermal neutron (220 cm/2.9 cm).
Figure 1.3 is a flow chart of Figure 1.2 showing four energy intervals during the
neutron life cycle. Labels indicate the dominant nuclear reaction in each interval.
In a critical reactor, only one neutron per fission completes the cycle, keeping the
number of neutrons constant.

Neutrons



Born at 2 MeV
Neutrons



Absorbed in Fuel
Neutrons


Thermalized
at 0.025 eV
Neutrons


Slowed to Range of
Resonances
Fast Neutrons

fast fission
Compound
Nucleus
Resonance
Energy
Neutrons
Parasitic Absorption

Thermal Neutrons
resonance capture
in U-238
fission
&
capture

Figure 1.3
Outline of the Neutron Life Cycle
Note the relative importance of the loss mechanisms while the neutron moves
through the lattice:
Notes:
CNSC 21
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

• resonance capture in U-238 accounts for about half of the losses;
• thermal neutron absorption in non-fuel material accounts for over one-third
of the losses;
• leakage (not shown in the figure) accounts for the remaining losses.

Careful choice of materials and their configuration keeps the net loss of neutrons
(fast, intermediate, and thermal) by leakage and capture to about 20% of all the
neutrons produced by fission. The remaining 80% are absorbed in the fuel as
thermal neutrons. With our choice of fuel, only about half of these cause fission,
just enough to return the same number as started the cycle
(2.5 neutrons/fission × 40% = 100%).
The parameter “k” (the neutron multiplication factor) characterizes the chain
reaction. The value k = 1 describes a reactor in which exactly one neutron from
each fission (on average) causes another fission. Such a reactor sustains a steady
fission rate and is said to be critical. This subsection develops an operationally
useful expression for k by following a group of neutrons around the flow chart and
looking at the production and loss mechanisms in each energy range. For
simplicity, the discussion begins in the interior of a very large reactor (the
“infinite” reactor) so that we can ignore leakage effects. The reactor comprises a
large number of cells, each assumed to contain exactly the same mix of fuel,
coolant, moderator, and other materials. By specifying identical cells in an infinite
reactor, we can assume there is no net transfer of neutrons between cells, reducing
the description to a single unit cell: a quadrant of Figure 1.2.
Notes:
CNSC 22
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Thermal Fissions



Born at 2 MeV
Neutrons



Absorbed in Fuel
Neutrons



Thermalized
Neutrons Reach



Resonance Energy
Fast Neutrons
Thermal Neutrons
N0 = 1000
858
930
1026
ε
Compound
Nucleus
Resonance
Energy
Neutrons
f
p
η
thermal
fission
capture
26: fast fission
96: resonance capture
72: parasitic absorption
FUEL
MODERATOR

Figure 1.4
The Neutron Life Cycle in the Unit Cell
(Numbers are for a Critical Infinite Reactor, k

= 1)
We will now follow a group of neutrons as they diffuse through the CANDU
lattice, that is, as the neutrons interacts with nuclei in the unit cell. Begin at the
top left of Figure 1.4. The initial number of neutrons is N
0
= 1000 fast neutrons
from thermal fission. Follow the heavy arrows around the closed loop as you read
through the following subsections.
Fast Fission
Starting with N
0
fission neutrons, the first thing that happens is that we actually
gain some neutrons, because fission neutrons leaving the fuel channel are
energetic enough for some of them to cause fast fission in U-238. The U-238 fast
fission cross-section is about half a barn for neutrons between 2 and 5 MeV. One
or two collisions with coolant molecules reduce the energy of a fission neutron
below the fast fission threshold (1.2 MeV). The designers calculate that 1000 fast
neutrons escaping the fuel channel generate 26 extra neutrons from fast fission.
The definition of the fast fission factor, denoted by the symbol ε (epsilon), is
fission thermal from neutrons of number
fission fast from neutrons of number
1+ = ε
For the design numbers in Figure 1.4,
ε = 1 + 26/1000 = 1.026
Notes:
CNSC 23
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The number of fast neutrons produced by all fissions (fast and thermal) in
the unit cell, starting with N
0
fast neutrons from thermal fission, is εN
0
(= 1026 in
our example).
The fast fission factor (ε) depends mainly on the relative amounts of fuel and
heavy water in the fuel channel. Voiding of coolant, which delays thermalization
until the neutrons have left the channel, is the only operational event that
significantly affects fast fission. Section 4 examines the effect of voiding the
coolant.
Resonance Capture in U-238
Capture of neutrons in the U-238 resonance peaks account for the biggest loss of
neutrons. Collisions with coolant nuclei can slow some fast neutrons to the
resonance energy range before the neutrons leave the channel. Fast neutrons that
escape into the moderator are not immune to resonance capture. Random
collisions in the moderator will return some neutrons to the channel of birth or
take them to nearby channels before thermalization is complete. Any resonance
energy neutron that enters the fuel faces almost certain capture.
The factor that accounts for resonance capture is the resonance escape probability
(p). This is the probability that a neutron will not undergo resonance capture in
U-238 while slowing down.
energy resonance reaching neutrons
resonances U in captured neutrons
1 p
238
92
− =

For the numbers in Figure 1.4
p = 1 – 96/1026 = 0.906
The number of neutrons reaching thermal energy in the unit cell, starting with εN
0

fast neutrons is pε N
0
(= 930 in our example).
A two-pronged approach limits losses by resonance capture:
• the moderator steps the neutrons quickly through the resonance energy
range to reduce the chance of an interaction.
• the fuel is “lumped” to shield its interior from resonance energy
neutrons (an effect described as “self-shielding”);
In effect, rapid slowing of the neutrons limits the flux of neutrons in the resonance
energy range; self-shielding (mentioned again in Section 4) limits the number of
Notes:
CNSC 24
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

U-238 targets. The combined effect is to limit the rate of resonance
capture, making criticality possible even with natural uranium fuel.
Section 4 examines three situations that significantly affect the resonance escape
probability:
• Increasing the fuel temperature, as occurs during a power increase,
quickly and dramatically increases resonance capture.
• Increasing the moderator temperature decreases moderator density.
More neutrons reach nearby channels while still in the resonance
energy range, so resonance captures increases.
• Voiding of the coolant reduces the number of resonance energy
neutrons available for capture, so resonance capture decreases.
Parasitic (Non-fuel) Absorption
Thermal neutrons in the unit cell are absorbed either in the fuel or in other core
materials. The thermal utilization (f) is the factor that accounts for non-fuel
thermal neutron absorption. It is the fraction of thermal neutrons absorbed by the
fuel compared to all the thermal neutrons absorbed in the cell.
( ) ( )
( ) ( ) ( ) ( ) fuel non fuel non fuel fuel
fuel fuel
a a
a
Σ φ + Σ φ
Σ φ
= f
φ(fuel) and φ(non-fuel) are, respectively, the average thermal neutron fluxes in the
fuel and away from the fuel. Accurate analysis requires taking into account the
variation in flux across the unit cell.

A simplified formula that assumes the flux is the same everywhere suits are
purposes here:
( )
( ) ( ) fuel non fuel
fuel
a a
a
⋅ Σ + Σ
Σ
= f
Using the numbers absorbed in fuel and non-fuel in Figure 1.4, we get
f = 858/930 = 0.923
Starting with pεN
0
thermal neutrons, the number of thermal neutrons absorbed in
the fuel is fpε N
0
(= 858 in our example).
Parasitic absorption is the second most important loss mechanism (after resonance
capture). The designers keep the non-fuel thermal neutron absorption to a
Notes:
CNSC 25
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

minimum by selecting low absorption materials for moderator, pressure
tubes, and other core components. The lattice pitch is kept small enough so that
thermal neutrons do not spend too much time diffusing, which would increase the
probability of their absorption in the moderator.
Thermal utilization is also the easiest factor to adjust for short-term control of the
fission rate. The designers have provided an array of neutron absorbing devices
intended for insertion or removal from the reactor to adjust neutron absorption.
Operating temperatures and coolant voiding affect f. Section 4 examines these
effects.
Fuel Absorption: Fission and Capture
All the neutrons remaining in the unit cell at this point are absorbed in the fuel.
The term “fuel” as used in this text includes all the content of the fuel bundle
(fissile material, U-238, fission products, Pu-240, etc.). Not every neutron
absorbed in fuel causes fission. For every neutron absorbed in equilibrium fuel,
fission returns 1.2 neutrons. This is the value of the reproduction factor denoted
by the symbol η (eta, pronounced “eight-ah”). Note carefully the distinction
between η, the number of fission neutrons per thermal neutron absorbed in the fuel
and ν, the number of neutrons produced per fission
) fuel ( ) fuel (
) fuel (
) fuel (
) fuel (
f
f
a
f
γ
Σ + Σ
Σ ν
=
Σ
Σ
× ν = η
The definition of η includes all fission neutrons, prompt and delayed. The fuel
cross sections are effective averages over all the contents of the fuel. The number
of neutrons per fission (ν) is an effective average over all the fissile isotopes.
Starting with fpεN
0
thermal neutrons, the number of neutrons returned to start the
next cycle is N = ηfpε N
0
(= 1000 in our example).
For the cycle in Figure 1.4:
η = 1000/858 = 1.166
The value of η required for criticality in a finite reactor is higher than the value
just given, to offset losses by neutron leakage. In a practical reactor, fuel
replacement keeps the inventory of U-235 higher and the inventory of fission
products lower, maintaining η near 1.2.
The reproduction factor depends on the fuel composition. Individual fuel bundles
change composition as they spend time in the neutron flux, but continual
replacement of high burnup bundles with fresh ones (on power refuelling) keeps
the core average composition the same from day to day.
Notes:
CNSC 26
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Operationally, the rate of replacing fuel, fuel temperature changes, and
coolant voiding all affect η. Section 5 examines fuel burnup effects and on power
refuelling. Section 4 discusses temperature effects and voiding.
The Lattice Cell Multiplication Factor
Summarizing the results above, we started the cycle with N
0
neutrons, and on
completing the cycle, the number of neutron returned was N = ηfpε N
0. We wi l l
introduce the symbol k

, (k-infinity, the lattice cell multiplication factor) such that
N = k

N
0
for the infinite lattice with no leakage. This gives:
k

= ηfpε
When there is a perfect steady-state balance between neutron production and loss
in the infinite lattice, k

= 1. This balance occurs in each energy range. The
production of fast neutrons (by thermal and fast fission) exactly balances the loss
of fast neutrons (by slowing the fast neutrons to lower energy). The production of
resonance energy neutrons (by slowing of fast neutrons) exactly balances the
losses of resonance energy neutrons (by resonance capture and thermalization).
The production of thermal neutrons (by thermalization) exactly balances the loss
of thermal neutrons (by absorption in fuel or other reactor core materials).
Criticality implies this sort of detailed balance.
Reactor engineers can calculate each of the four factors if they know the
composition and geometry of the core. This allows them to compare and optimize
configurations and materials. The numerical example carried throughout our
discussion used design values of the factors for a critical infinite lattice, but the
engineering formulas for calculating the factors do not make any prior
assumptions about criticality. The four-factors and k

can be evaluated for any
mixture of materials, not just for k

= 1.
1.4 THE FINITE REACTOR
Now we will extend the idea of neutron balance across a finite core, where each
cell has a different composition and different flux. In the reactor, the random
motion of individual neutrons results in an overall movement of neutrons from
regions of higher neutron density to regions of lower density. (In the jargon, net
neutron transfer depends on the flux gradient). In the finite reactor, the neutron
balance in each lattice cell now includes neutrons that transfer across cell
boundaries. For example, production of thermal neutrons in a cell would include
production by thermalization, as before. It could also include the transfer of
thermal neutrons into the cell from a neighbouring cell. Similarly, losses from the
cell, in addition to those discussed for the infinite reactor, could include transfer of
neutrons out of the cell.
The principle of criticality is the same; a steady-state critical core exists when
neutron production balances neutron losses in each cell for neutrons in each
Notes:
CNSC 27
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

energy range. In the finite reactor, we allow for fuel bundles with
different composition in each cell. There are also some cells with extra absorption
because some bits of reactor hardware (adjuster rods, measuring instruments etc.)
pass through them. There are cells in the reflector where there is no neutron
production by fission, only by transfer from neighbouring cells. At the boundary,
neutrons transfer into the shielding, but we assume no neutrons come back.
For the finite core, each cell links to each neighbouring cell (and indirectly to
every other cell in the reactor) because neutrons lost from one cell (in any energy
range) become neutron production in a neighbouring cell. The net transfer from a
cell is called cell leakage. It could be positive (transfer out of the cell), or negative
(transfer in). To achieve a neutron balance within one cell means balancing
neutrons in all cells. Anything that affects an individual cell inevitably affects, at
least to some degree, every other cell.
The effective multiplication factor (k) accounts for possible net imbalance across
the entire finite core. Quite clearly, we cannot put together an arbitrary mixture of
fuel, moderator, reactor hardware etc. and expect removal and production will be
equal. When production is greater than loss, k > 1 (the core is supercritical) and
when production is less than loss, k < 1 (the core is subcritical).
Criticality occurs for a single value k = 1 across the core. This involves bringing
neutrons in every energy group into balance at every point in the reactor
simultaneously. For the situation of perfect balance (the steady-state critical
reactor) mathematicians are able to determine steady state fast and thermal neutron
fluxes by solving the complicated set of interconnected equations. Section 8
discusses flux shape.
There is no steady state solution for flux when k ≠ 1. With k > 1 flux is
increasing. For k < 1 flux is decreasing—if there are no neutron sources present.
Section 2 introduces power change in the supercritical core and Section 3
introduces the behaviour of the subcritical core.
The Six Factor Formula
The composition of each cell in the finite reactor determines the four factors of k


for each cell. By this we mean the values the four factors would have in an
infinite reactor with the composition of the particular cell. Using effective average
factors across the whole core, we can write the effective multiplication factor:
k = ηfpε Λ
t
Λ
f

The factor Λ
t
(lambda-t) accounts for the transfer of thermal neutrons out of the
cells at the reactor boundary. It is the probability that a thermal neutron will not
leak out of the reactor, so its name is thermal non-leakage probability.
Notes:
CNSC 28
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The fast non-leakage probability denoted by Λ
f
(lambda-f) is the factor
that account for the transfer of fast neutrons out of the cells at the reactor
boundary. It is the probability that a fast neutron will not leak from the reactor.
The fast and thermal non-leakage probabilities depend on:
• How far the neutrons wander as they diffuse,
(i.e., on the number of collisions and the path lengths);
• The shape and size of the reactor;
• The reflector.
The first two of these are related. In a small core, too many neutrons will reach
the edge and leak out. It is not possible to achieve criticality below a particular
size, known as the critical size. The reflector returns some of the leakage neutrons
to the core, making criticality possible with a smaller core size.
Reactivity
When k = 1 the neutron flux and the power level remain constant.
1
It is important
to note that criticality depends on neutron balance, not on power level; saying that
a reactor is critical is not the same as saying it is at high power.
Under normal operating conditions a reactor operates close to criticality (that is, k
is nearly equal to one). It is more convenient under these circumstances to talk in
terms of the amount by which k differs from one than it is to keep quoting the
value of k itself.

The effective multiplication constant is defined in terms of a detailed neutron
balance everywhere in the core:
(neutron removal) =
k
1
(neutron production)
The balance exists for k = 1. The amount by which (1/k) differs from one defines
reactivity.
k
1 k
k
1
1 activity Re

= − =
Reactivity is a measure of `the imbalance between neutron production and removal
across the entire core.

1
Steady power occurs with k < 1 when sources are present. We ignore sources in this section.
Notes:
CNSC 29
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Because we usually deal with values of k that are close to one, the above
expression is nearly equal to k-1 and the following is a satisfactory approximation:
k 1 k activity Re ∆ = − ≈
In this course, we usually take this absolute difference between k and one as the
reactivity.
Reactor regulation is achieved with very small reactivities, so numerical values of
reactivity are usually written in a notation based on the metric prefix “milli-”, the
milli-k (mk). The following example illustrates the use of the mk to express
reactivity = ∆k:
Suppose we have k = 1.002
then, ∆k = k-1 = 1.002 - 1 = 0.002
In this case, ∆k is equal to 2 parts in a thousand, which we call 2 mk:
∆k = 2 mk (Notice that ∆k/k = 1.996 mk)
Conversely, for ∆k = -1 mk, k = 0.999.
The notation ∆k for reactivity is easily misused. The notation “∆” is used
generally to mean any difference between two values of a parameter, but in the
equation for reactivity, it is the difference from a fixed reference point, the critical
core. (It is k - 1, not k
2
– k
1
).
The term reactivity worth expresses the difference between two different
reactivities, and the term excess core reactivity describes a change in core
reactivity from a defined configuration. The following are some common
expressions associated with reactivity.
Reactivity worth describes the change in reactivity a device could introduce into
the core, whether or not the core is critical. For example, the reactivity worth of
the liquid zones between full and empty is about 6 mk.
Excess reactivity is the difference in reactivity between the actual core and the
nominal core. The nominal core is a core configuration, defined for each station,
in which each reactivity device is in its “normal” operating state. A typical
nominal core would have the adjuster rods fully inserted, control absorbers and
shutoff rods poised, liquid zone compartments 40% full, with reactor grade
moderator (and coolant), free of dissolved poisons, and normal system
temperatures.
Control reactivity is the total reactivity worth of the control devices compared to
the nominal configuration. To keep the core critical, the control reactivity and the
core excess reactivity must be equal and opposite.
Notes:
CNSC 30
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

For example, suppose the core is critical (k = 1, ∆k = 0) in the nominal
configuration and the rate of fuelling is increased to allow the fuelling machines to
be removed from service. The regulating system holds power by raising the liquid
zone levels to keep k = 1 during fuelling. Following fuelling the core has excess
reactivity, even though ∆k = 0. The difference in zone level from the nominal
40% measures the control reactivity, which exactly offsets the excess core
reactivity.
Notes:
CNSC 31
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

RESPONSE OF THE CRITICAL REACTOR TO A
REACTIVITY CHANGE
2.0 INTRODUCTION
This section deals with a topic of direct operational importance—the way the
reactor responds to an imposed change in the effective multiplication constant,
that is, a change in core reactivity, which measures the difference between a
critical core (k = 1) and a core where neutron production and removal do not
balance.
The neutron flux increases exponentially when the reactor operates with a
positive reactivity (k > 1). The reactor period, which governs this rate of rise,
depends on the reactivity and on the neutron lifetimes in the system. The long
lifetimes of the delayed neutrons are crucially important in limiting the rate of
power rise, but their effect becomes less important as the size of the reactivity
insertion increases.
• For a small reactivity insertion in a reactor core, the sort used in
normal regulation, the exponential rise is well controlled.
• Reactor power increases very quickly for a fraction of a second
following a larger reactivity insertion. After this prompt jump, the
delayed neutrons control a slower stable power rise.
• Still larger reactivity insertions make the reactor prompt critical,
causing an uncontrollably fast rise of the sort that occurred at
Chernobyl.
The reactor makes a smooth, gradual transition between operating regimes, but
the accurate description of reactor response gets more complicated as the
reactivity insertions get larger. Consequently, this section uses different models
to account for the observed reactor behaviour.
Power reduction (a prompt drop followed by a gradual power rundown) is part
of normal operation, but operating staff should never see a prompt jump (or
prompt criticality). Reactivity insertions must be limited in size (by design and
operation) to limit reactor power increase. Station staff must ensure that
changes to systems or operating practice do not violate the built in constraints.
2.1 EXPONENTIAL POWER RISE
December 2, 1942 saw the first controlled, self-sustaining nuclear chain
reaction. Enrico Fermi was in charge. Here is an eyewitness account:
“But suddenly, his whole face broke into a broad smile.” ‘The reaction is self
sustaining,’ he announced quietly, happily. ‘The curve is exponential.’”
Notes:
CNSC 32
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Exponential power rise characterizes a supercritical nuclear reactor; power rises
as follows:
τ
=
t
e P ) t ( P
0

P
0
is the power at time t = 0, and power in this equation (and throughout this
section) is neutron power; it does not include decay heat. Section 7 will
distinguish between neutron power and reactor thermal power.
The parameter that characterizes the rate of power rise is the reactor period, τ
(Greek tau). The reactor period, by definition, is the time it takes power to rise
by a factor e, the base for natural logarithms (e ≈ 2.718…). A long reactor
period corresponds to a slow power rise, a short period to a fast power rise.
Dividing the exponential relationship by P
0
and taking the natural logarithm
gives the equivalent relationship:
τ =
|
.
|

\
|
/ t
P
P
0
ln
Figure 2.1 shows this linear power increase with time. We also describe this
behaviour by saying that the rate of change of logarithmic power (log rate in
station jargon) is constant.
1
10
100
1000
10000
0 300 600 900 1200 1500 1800
time (s)
l
n
(
P
/
P
0
)
P
0
P(t)
t

Figure 2.1
Logarithmic Power Increase for an Exponential Power Rise
Log Rate

Notes:
CNSC 33
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Using the expression above, the average rate of change of the logarithm of
power over the time interval t is
( ) ( )
τ
=

=


=
1
t
P P
t
P
rate log
0
ln ln ln

Using the exponential equation, the following relationships give the
instantaneous rate of change
2
:
τ
= = =
1
dt
dP
P
1
dt
P d
rate log
ln

By either approach, we conclude that
log rate = 1/τ
Whenever the time rate of change of the logarithm of some variable equals a
constant, (or, equivalently, the time rate of change of the variable is proportional
to the variable) integration of the equation demonstrates exponential time
dependence.
The log rate represents a fractional increase per second. The rate log meters on
CANDU control room panels give the rate log in units of percent present power
per second (% P.P. s
-1
)
2.2 CORRECTIONS TO EXPONENTIAL REACTOR RESPONSE
The simple exponential behaviour described in the last section applies only to a
slightly supercritical reactor at low power. When the reactor is operating at
high power, feedback effects complicate the response to an imposed reactivity.
For example, if the control system increases reactivity, power starts to rise,
causing changes in the temperatures of various reactor components. This in turn
affects the reactivity of the reactor. In this section, we assume that the control
system inserts a reactivity ∆k, and that nothing changes this original value. This
is exactly the kind of behaviour observed during low power operation where
feedback reactivity effects do not occur, or are small enough we can ignore
them. Later sections examine feedback.
The second deviation from simple exponential behaviour is the response of the
reactor to a rapid insertion of a larger reactivity. In practice, an important
feature of CANDU operation is that during normal operation the automatic
regulating system adjusts reactivity. The computer software cycles through a

2
the relation
dt
dP
P dt
P ln d 1
= is a calculus identity.
Notes:
CNSC 34
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

sequence of tasks, and, once per cycle, if necessary, it initiates a small step
change to move the measured power level towards the requested power level.
The likely variation in reactivity is a ramp change made of a sequence of very
small step changes. However, we will illustrate reactor response with a single
step change, which is easier to examine than a sequence of steps.
The Prompt Jump Approximation
Many texts introduce a model of nuclear dynamics that produces a set of
differential equations with an analytic solution. In spite of the relative
simplicity of the model, it quite accurately mimics more detailed modeling. The
“two group” reactor dynamics theory models the neutron population with a
large group of prompt neutrons and a single, small group of delayed neutrons.
For a critical reactor with initial steady state power P
0
, the power as a function
of time, P(t), after an addition of a step reactivity making k > 1, is:
(
¸
(

¸

β
ρ

¦
)
¦
`
¹
¦
¹
¦
´
¦
|
|
.
|

\
|
β
ρ

=
τ

τ
1
e e
P
) t ( P
C
t
t
0

where
k
1
1
k
1 k
− =

= ρ is the core reactivity
and
|
|
.
|

\
|
β
ρ
λ
|
|
.
|

\
|
β
ρ

= τ
1
is the reactor period.

|
|
.
|

\
|
β
ρ

|
|
.
|

\
|
β
= τ
1
C
l
, where l ≈ 10
-3
s is the time for one prompt neutron
cycle.
The “transient time constant” (τ
C
), is very much smaller than the reactor period,
except when ρ ≈ β; but the equation is not valid for ρ ≈ β anyway. This
equation is a good approximation if the final reactivity (ρ), is not close to the
value of β. For small reactivity insertions k ≈ 1, as it must be for control of a
critical core, so ρ ≈ ∆k.
In the first formula above, the final term in the curly bracket is a transient term
that gives the time dependence of the initial power rise during the first fraction
Notes:
CNSC 35
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

of a second. Ignoring the time variation in the first fraction of a second
and substituting ∆k for the reactivity gives us the prompt jump approximation:
τ
(
¸
(

¸

∆ − β
β
=
t
e P
k
) t ( P
0

The reactor period is given by
|
.
|

\
|
∆ λ
∆ − β
= τ
k
k

Figure 2.2 shows this power increase for a reactivity insertion of ∆k = 0.5 mk.
Later in this section, we will use a numerical model to show why the reactor
behaves this way.
The long-term power increase in Figure 2.2 is generally similar to what we saw
in section 2.1. However, on inserting the step ∆k = 0.5 mk at time t = 0, the first
thing seen is a rapid power increase—just over 10% (in this case).

P o w e r R i s e
1 .0 0
1 .2 0
1 .4 0
1 .6 0
1 .8 0
2 .0 0
2 .2 0
2 .4 0
2 .6 0
2 .8 0
3 .0 0
0 3 0 6 0 90 1 2 0 15 0
T im e ( se co n d s )
P
(
t
)
/
P

Figure 2.2
Power Rise after a Moderate Reactivity Insertion (∆k = 0.5 mk)
This initial fast rise is the prompt jump, a completely new feature. The more
gradual exponential increase following the prompt jump, the stable rise, is
nearly the same as the exponential rise in section 2.1.
In the figure, the size of the prompt jump is 11 . 1
0005 . 0 0050 . 0
0050 . 0
k
=

=
(
¸
(

¸

∆ − β
β

The size of the prompt jump depends on the size of the reactivity insertion (or,
more precisely, on the ratio of ∆k/β). Notice that the prompt jump is negligible
when ∆k << β.
Notes:
CNSC 36
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

because 1
k

(
¸
(

¸

∆ − β
β
for ∆k << β
In the figure, the reactor period is s 5 . 112
0005 . 0 08 . 0
0005 . 0 0050 . 0
k
k
=
×

=
|
.
|

\
|
∆ λ
∆ − β
= τ
The reactor period also simplifies for small ∆k.
k k
k
∆ λ
β

|
.
|

\
|
∆ λ
∆ − β
= τ for ∆k << β
In other words, when reactivity (∆k) is a very small fraction of the delayed
neutron fraction (β), the prompt jump formulas revert to a simple exponential
rise. The exponential power rise calculated with this formula for τ agrees with
actual power rise for small reactivity additions such as those used by the
CANDU regulating system (∆k < 0.1 mk). It is less accurate for step additions
a little larger than this. The prompt jump approximation is adequate for larger
additions as well as for small reactivity additions.
2.3 THE EFFECT OF DELAYED NEUTRONS
The reactor period, which governs the rate of stable rise, is proportional to the
delayed neutron lifetime (1/λ), and also depends on the ratio of the reactivity to
the delayed neutron fraction (∆k/β). Reactivity devices can adjust reactivity.
Parameters β and λ change with fuel composition, but are very nearly constant
in an equilibrium fuelled CANDU. The reasons why reactor control depends
crucially on the delayed neutrons is examined next.
The Average Lifetime Model
The simplest model of reactor behaviour (the average lifetime model) ignores
the vast lifetime difference between prompt and delayed neutrons. It assumes a
neutron generation time (time from one fission to the next) that is a weighted
average of the prompt and delayed neutron lifetimes. For equilibrium CANDU
fuel, the generation time is approximately:
99.5% × 10
-3
s + 0.5% × 12.5 s = 0.000995 s + 0.0625 s = 0.063 s
The prompt neutrons have such a short lifetime compared to the delayed
neutrons that the generation time is calculated accurately enough for this crude
model as the product of delayed neutron fraction (β) and the average delayed
neutron lifetime (1/λ).
Generation time ≈ β/λ ≈ 0.06 s
Notes:
CNSC 37
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

In this model, when k > 1, power rises from generation to generation, and
the generation time governs the rate of rise. For a short neutron generation
time, the increase occurs many times in a second. For a longer generation time,
there are fewer cycles in one second and the rate of power increase is less. If all
the neutrons were prompt fission neutrons, the generation time would be very
short ( l ≈ 10
-3
s) and the rate of rise would be uncontrollably fast.
The power rise during one generation (∆P) is
∆P = kP – P = P(k-1) = P∆k.
The time for this power increase is the average generation time, β/λ, so the
average rate of change over one cycle is
( )
P
k k P
t
P
β
∆ λ
=
λ β

=



Equating this average increase in a generation with the instantaneous rate of
change, dP/dt, yields the following equation:
τ
=
β
∆ λ
= =
1 k
rate log
dt
dP
P
1
so
k ∆ λ
β
= τ
This constant rate log confirms a simple exponential power rise that depends on
the generation time, which is determined almost entirely by the delayed
neutrons. For this model, the effect of the delayed neutrons is to make the
generation time (and the reactor period) much longer than it would be if there
were only prompt neutrons.
The Two Group Model
The most noticeable difference between the prompt jump approximation and the
neutron average-lifetime approximation is the prompt jump. Treating all the
neutrons alike is an oversimplification that fails to predict the power change
immediately following the change in reactivity. The prompt power rise is the
immediate response of the 99.5% of the neutrons with one-millisecond
lifetimes. The prompt power rise ends quickly and a slower, exponential rise
continues, controlled by the delayed neutrons.
To understand this behaviour, we must abandon the unrealistic model of a
neutron generation based on an average neutron lifetime and consider the time
behaviour of the prompt and delayed neutrons separately. Figure 2.3 illustrates
how we will do this and introduces the notation.
Notes:
CNSC 38
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Figure 2.3
A Model for Separate Prompt and Delayed Neutrons
The figure shows one prompt neutron cycle ( l = 1 ms). The multiplication
factor is k, so N neutrons of one “generation” give rise to a total of kN in the
next “generation”. The majority of the neutrons are prompt, so the prompt
fraction, (1 - β) × kN neutrons, appear immediately. The delayed fraction,
β × kN, are added to a “bank” of delayed neutron precursors, to be released
some time later. The delayed neutrons feed into the fast cycle at a rate
controlled by the delayed neutron lifetime and the bank size.
For a steady-state critical reactor, the precursor bank is at an equilibrium
concentration where the decay from the bank equals the input. On a step
increase in reactivity, the prompt population increases quickly because of an
increase in (1 - β) k. The delayed neutron contribution increases slowly, as it
depends on the precursor bank.
You may wish to demonstrate this behaviour with a numerical example. Take
N = 1,000 initially, and to make the computations easier, use a greatly
exaggerated delayed neutron fraction, β = 10% instead of 0.5% and a reactivity
insertion ∆k = 0.050 (in other words, k goes from 1.000 to 1.050). This
simplifies the arithmetic, as there is a significant change from one prompt cycle
to the next, but in no way alters the qualitative arguments.
Before the reactivity insertion, N = 1000, and in each cycle 100 of these are
stored in the precursor bank, exactly balanced by 100 that enter the cycle by
precursor decay. Now, since k is greater than one, the total number of neutrons
produced by fission rises to kN = 1050. Ten percent of these are in the form of
delayed neutron precursors, so 105 neutrons are “stored” in the precursor bank,
to be released later. The prompt neutrons number 945, to which we add 100
neutrons from the precursor bank (stored earlier).
Notes:
CNSC 39
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This gives a total of 1045 in the second generation.
If you repeat this calculation for a few more cycles, you should notice two
things:
Although the input to the precursor bank continues to increase, the output stays
constant as the neutrons coming from it reflect a precursor concentration
accumulated before the change in reactivity took place. (Each cycle is only
1 ms compared to the “delay time” of 12.5 seconds).
Although the neutron density rises rapidly, the rate of rise falls off (increases of
45, 42, and 40 in successive generations). The reason it falls off is that, at each
generation the precursor bank progressively stores more fission neutrons rather
than immediately releasing them into the chain reaction.
If you persist and follow, say, fifty cycles, (0.05 seconds after the reactivity
insertion) you will see that the increase from one cycle to the next is quite small,
and the number of potential neutrons going into the precursor bank each cycle is
quite high.
The result is that the rate of increase of the neutron density falls off to zero after
a few hundred generations. If we assume that there is no increase in the output
of the precursor bank, the situation after 1,000 generations (1 s) would be stable
as shown in Figure 2.4.

Figure 2.4.
Eventual End of the Prompt Jump (with k= 1.05 and β = 1.10)
By this time, the precursor bank has increased significantly and the number of
decays in each cycle will gradually increase from the initial 100 to 101, then
102 etc. The arrival of these delayed neutrons keeps the power increasing.
FISSION
kN=1909
1818 neutrons
Precursor Bank
1718
prompt
191
delayed
100 delayed
1818 neutrons
1718 + 100
Notes:
CNSC 40
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

In summary, a step increase in ∆k makes k > 1, but (1-β) × k < 1. (If no
neutrons came out of the precursor bank, the reactor would be subcritical).
Prompt multiplication causes the neutron population to rise quickly, but only
because the neutron bank continues to “top up” the neutron population at the
same rate as before the reactivity addition. Extra precursor decays from the
accumulation in the precursor bank drive the stable rise that follows the prompt
jump.
Example
A few seconds after the insertion of 0.5 mk of positive reactivity (i.e.
∆k = 0.0005), reactor power is observed to rise 100 fold (e.g. from 10
-4
F.P. to
10
-2
F.P.). Compare the times of the power rise:
a) For a real reactor with delayed neutrons,
b) If all the neutrons could be prompt.
In each case the power rises exponentially, given by
τ
=
t
e P ) t ( P
0

a) The period for a real reactor is
τ = (β - ∆k)/λ∆k = (0.0050 – 0.0005)/(0.08 s
-1
× 0.0005) =112.5 s
This gives t = τ × ln(P/P
0
)
= 112.5 s × ln(100) = 518 s ≈ 8½ minutes.
b) If all the neutrons were prompt, the period would be
τ = ℓ/∆k = 0.001 s/0.0005 = 2 s
This gives t = τ × ln(P/P
0
)
= 2 s × ln(100) ≈ 10 s.
This is almost sixty times faster than actually occurs. Comparing the rates of
power increase:
a) Rate log = 1/τ = 1/ 112.5 ≈ 0.9% P.P./s
b) Rate log = 1/τ = 1/2 = 50% P.P./s
A rate of 0.9% P.P. s
-1
is a fast power-manoeuvring rate, used only at low
power. A 50% P.P. s
-1
rise is much too fast to control.
2.4 PROMPT CRITICALITY
To this point, we have restricted the reactivity addition ∆k to the range ∆k << β,
or at least to values not very close to β. What happens if we insert a large
positive step-change of reactivity, greater than β? Go back to our example,
where β = 0.1, and insert a reactivity step ∆k = +0.15 (k = 1.150).
Notes:
CNSC 41
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Figure 2.5 shows that starting with 1,000 neutrons in generation 1, the
number of prompt fission neutrons in generation 2 is 1035.

Figure 2.5
Prompt Supercritical
The reactor becomes supercritical on prompt neutrons alone, without having to
wait for the delayed neutrons. A reactor with an effective multiplication factor
large enough to make the prompt neutron fraction increase enough to make the
reactor critical without waiting for the prompt neutrons is prompt critical.
The condition for prompt criticality is (1-β) × kN = N
Dividing through by kN and isolating β on one side of the equation gives
β = 1 – 1/k = (k-1)/k (= reactivity)
or ∆k ≈ β
For the equilibrium core, the value of β is approximately 0.005, so prompt
critical occurs for ∆k ≈ 5 mk. Under these circumstances, the exponential
power rise from the fast neutrons alone is similar to that in the example where
we assumed all the neutrons were prompt. None of our formulas are valid for
∆k ≈ β, but accurate numerical analysis shows the reactor period is a little less
than 1 second for a prompt critical CANDU reactor, making the rate of power
increase (rate log = 1/τ) greater than 100% P.P/s.
There is no abrupt change in reactor behaviour as we move through the prompt
critical condition. As reactivity ∆k → β, the reactor becomes less and less
dependent on the delayed neutrons so that reactor period (τ) decreases smoothly
to shorter and shorter values.
FISSION
kN = 1150
1000 neutrons
(Generation 1)
Precursor
Bank
(1-β) × kN
0.9 × 1150
1035 prompt
β × kN
115 delayed
100 delayed
1135
(Generation 2)
1035 + 100
Notes:
CNSC 42
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The rate of reactor power increase is unacceptable for reactivity insertions
well below ∆k = β.
Automatic reactor shutdown occurs when instruments detect a high rate of
power increase: reactor stepback drops control absorber rods into the core, a trip
on shutdown system one (SDS1) drops shutoff rods into the core, or a shutdown
system two (SDS2) trip injects neutron-absorbing poisons into the moderator.
These systems trigger at rates corresponding to the following reactor periods:
τ = 12.5s (rate log 8% P.P. s
-1
) for stepback,
τ = 10 s (rate log = 10% P.P. s
-1
) for SDS1,
τ = 6.7 s (rate log = 15% P.P. s
-1
) for SDS2.
2.5 POWER RUNDOWN: THE PROMPT DROP
We can use the equations describing the prompt jump to estimate power
decrease following a large negative reactivity insertion. In this case, we have a
prompt drop as the prompt neutron population collapses, followed by a stable
negative period. With a negative reactivity insertion, the prompt population
multiplication factor (1 - β) k is much less than 1 so the prompt population
drops. The precursors bank gradually decays as there are not enough fissions to
replenish it.
The prompt “jump” approximation gives the ratio of power P immediately after
the step insertion of reactivity to the original level P
0
, but in this case, ∆k is
negative. For example, a full stepback may introduce approximately 15 mk of
negative reactivity. If ∆k = -15 mk is injected into a critical reactor, the power
immediately after the step change is,
0 0 0
P 25 . 0
) 015 . 0 ( 005 . 0
005 . 0
P
k
P P =
− −
=
∆ − β
β
=
Following the prompt drop, the power settles down to decaying with a negative
period. The equation for the reactor period also works for negative reactivity.
For ∆k = -15 mk
k
k
∆ λ
∆ − β
= τ gives
s 17
) 015 . 0 ( 08 . 0
) 015 . 0 ( 005 . 0
− =
− ×
− −
= τ


Notes:
CNSC 43
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This represents a decreasing exponential that looks like decay. The power
continues to drop with a decay constant of approximately 1/17 = 0.06 s
-1
.
If |∆k| is >> β, the period becomes approximately
λ
− →
∆ λ
∆ − β
= τ
1
k
k

Thus, the decay constant of the precursors determines the stable period for a
large negative reactivity insertion. In this case, power decays following the
prompt drop with a decay constant of approximately 1/12.5 = 0.08 s
-1
.
For large negative ∆k, the approximation becomes increasingly inaccurate
because it models all the delayed neutrons as though they have the same decay
constant. Consider the fact that there are actually six precursor groups, the
longest-lived having a half-life of 54 s. The power, after the shorter-lived
groups have decayed, will decrease with a period (τ) of about 80 s. (τ = 1/λ and
λ = 0.693/t
1/2
).
As Section 7 describes, the rate of power decrease eventually slows even further
because of the presence of photoneutrons, which have longer half-lives than the
delayed neutrons.
A final cautionary note: remember that this section deals only with the neutron
power. Following a reactor shut down from high power, the thermal power is
not at all proportional to the neutron flux, due to the heat contributed by the
accumulated fission products.

Notes:
CNSC 45
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

RESPONSIVENESS OF THE SUBCRITICAL REACTOR
3.0 INTRODUCTION
CANDU reactors have two neutron sources (not including neutrons from induced
fission). These are photoneutrons and neutrons from spontaneous fission. In this
section, we show that the fission process amplifies the source neutrons by a factor
(the subcritical multiplication factor) that is inversely proportional to the
reactivity.
The nuclear instruments can measure observable flux in a subcritical core, even
when the source is quite small, because there is subcritical multiplication of source
neutrons. The observed flux increases as k increases in the subcritical core
because the source amplification is inversely proportional to reactivity. When
deeply subcritical, very large changes in reactivity produce little noticeable change
in flux. In contrast, an almost critical reactor behaves, for all practical purposes,
nearly like a critical reactor. The essential difference is that for k > 1, power
continues to rise; for a reactivity increase that leaves k < 1 the power stops rising
at a level that depends on reactivity.
The section also examines the stabilization time, the time for the power level to
establish a new value after a step change in reactivity. The stabilization time gets
longer the closer the reactor is to being critical.
Operationally, we can determine the reactivity of the shutdown reactor by
inserting or withdrawing a known reactivity worth, and measuring the
corresponding change in indicated power. In particular, a power doubling occurs
in a subcritical reactor whenever we reduce the shutdown reactivity by half.
Procedures use this behaviour to approach criticality cautiously during startup, and
to test whether or not a low power reactor is critical.
3.1 NEUTRON FLUX IN A “SHUT DOWN” REACTOR
The observed neutron power (flux) in a subcritical reactor with sources present
does not drop quickly to zero. The observed neutron power is considerably higher
than can be accounted for by the size of the neutron source itself. To explain the
phenomenon, consider a reactor held deeply subcritical by neutron absorbing
poisons dissolved in the moderator. Now suppose we inject a single pulse of
neutrons into the core. Most of the neutrons are absorbed without causing fission
but, inevitably, some cause fission and generate additional neutrons. (The
effective multiplication factor of the “shut down” reactor is much less than one,
but it is not zero.)
Introducing a pulse of neutrons does not initiate a self-sustaining chain reaction;
(less than one neutron per fission survives). However, if a neutron source injects a
steady trickle of neutrons, the instruments register more flux than just the source
Notes:
CNSC 46
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

flux. Moreover, as k gets closer to k = 1, more neutrons survive from each
fission that occurs and the flux increases. The flux would drop quickly to zero
without source neutrons in the reactor, but when there is a source, there are always
more neutrons than just those from the source.
The next two subsections address, respectively, the natural sources of neutrons in a
CANDU reactor, and the amplification effect due to the reactor.
Neutron Sources
The term “source neutrons” refers to neutrons other than prompt or delayed fission
neutrons. A neutron source injects a steady supply of neutrons, independent of
power level, temperature, or any controllable parameter. A CANDU has two
“built-in” neutron sources, one arising from the nature of the fuel, and the other
from the presence of heavy water.
1. Spontaneous fission neutrons
Neutrons from spontaneous fission come mainly from U-238, which constitutes
most of the fuel. The contribution to reactor power from fissions caused by the
spontaneous fission neutrons is constant, at about 10
-12
% of full power. This low-
level source is important only at first startup (or after a prolonged shutdown) when
there are essentially no other neutrons.
NOTE: Photoneutrons
High-energy gamma rays (> 2.2 MeV) emitted during fission product decay
generate photoneutrons by interactions with deuterium nuclei. The photoneutron
source strength depends on the presence of energetic gamma rays, which in turn
depends on how long the reactor has been operating. Following prolonged
operation at significant power levels (>10% full power), the decay gamma flux is
proportional to the power. The photoneutron source is typically about 0.03% of
the power level—less than 10% of the delayed neutron fraction. This source
becomes important following the disappearance of the delayed neutrons after
shutdown.

Notes:
CNSC 47
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Neutrons Contribution
Prompt Neutrons 99,470,000,000,000
Delayed Neutrons 500,000,000,000
Photoneutrons 30,000,000,000
Spontaneous Fission 1
Total (10
14
n cm
-2
s
-1
) 100,000,000,000,000
Table 3.1
Relative Neutron Strengths
Table 3.1 shows the relative importance of source neutrons by showing the
makeup of the full power flux.
Source Neutron Time Dependence
Spontaneous fission produces a small, constant supply of neutrons (about 130
neutrons per second per CANDU bundle).
Following shutdown, the photoneutron source decreases with the decay of fission
products that produce high-energy gamma rays. Power from fissions caused by
photoneutrons is typically close to 5 × 10
-5
% one day after shutdown. (The
precise value depends on the operating power in the weeks before shutdown.)
After the initial rapid drop, the photoneutron source drops slowly, decreasing a
little each day. The longest-lived fission product with gamma ray energy above
the threshold is Ba-140, which has a half-life of 12.75 days; thus, the photoneutron
source persists for many weeks after shutdown.
This course follows the usual convention of treating neutrons from delayed
neutron precursors as delayed fission neutrons, and neutrons from
photodisintegration of deuterium as source neutrons. The distinction rests on the
practical operational effects that result from the different half-lives of these kinds
of neutrons.
The longest-lived delayed neutron precursor has a 54-second half-life. When the
fission rate changes, it takes a few minutes for the delayed neutron precursor bank
to come back into equilibrium. Its slow change controls the rate of change (in
both the subcritical and the critical core) of the current fission rate. The current
fission rate determines the size of the equilibrium precursor bank.
Photoneutrons, in contrast, have much longer precursor lifetimes. Weeks after a
shutdown, fission products from power operation in the weeks before shutdown
continue to generate photoneutrons. Following a power change, it takes weeks for
the photoneutron fraction to reach its new equilibrium strength.

Notes:
CNSC 48
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

For neutron balance in a critical steady-state core (Section 1), it is
convenient to lump the photoneutrons with the delayed neutrons—they act like
delayed neutrons with long lifetimes. Dynamic reactor behaviour (Section 2)
depends on a delayed neutron “delay tank”—a neutron source that changes slowly
(over many prompt neutron cycles) following a reactivity insertion. The
distinction between source neutrons and delayed fission neutrons depends on
whether or not the neutron supply is nearly constant over the time interval
considered. Operationally, things like power ramps take minutes or hours. From
an operational perspective, that makes neutrons from delayed neutron precursors
delayed fission neutrons, while photoneutrons are source neutrons.
Source Multiplication in a Subcritical Reactor
The previous sub-section stated that the contribution to reactor power from
fissions caused by the spontaneous fission neutrons is about 10
-12
% of full power,
and the contribution from photoneutrons is 0.3% of full power. These values are
for fissions caused directly by the source neutrons themselves; that is, we ignored
the fact that these fissions give rise to additional fission neutrons that, in turn, may
cause further fission. The subcritical reactor acts as a multiplier of source
neutrons, so that the actual power generated because a source is present is much
greater than would be produced by the source neutrons alone. Source
amplification in the subcritical reactor depends on its reactivity, as shown below.
Assume that a neutron source is inserting S
0
neutrons in each generation into a
subcritical reactor whose effective multiplication factor, k, is less than one.
Imagine that you could label the neutrons so that you could identify the origin of
each neutron in the reactor. In the neutron population present in the core now,
there are S
0
neutrons emitted by the source during the current generation.
During the immediately preceding generation, the source emitted S
0
neutrons and
some of these source neutrons caused fission, producing a contribution of kS
0

neutrons to the present population.
The total number of neutrons in the neutron population from this and the
immediately preceding generation is therefore:
S
0
+ kS
0
(kS
0
is smaller than S
0
).
Similarly, one generation earlier, the source emitted S
0
neutrons, generating kS
0

neutrons in the immediately preceding generation, resulting in k
2
S
0
neutrons in the
present generation. Looking back, the contribution of three generations of source
neutrons and the fissions they caused contributes
S
0
+ kS
0
+ k
2
S
0



Notes:
CNSC 49
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

If we extend this argument indefinitely (working backwards a large
number of generations), we see that the total neutron population (S

) is made up as
follows
S

= S
0
+ kS
0
+ k
2
S
0
+ k
3
S
0
+ ...
= (1 + k + k
2
+ k
3
+ ...) S
0

The series in brackets has an infinite number of terms, but a high-school algebra
trick yields a finite sum. For k < 1

0
S
k 1
1
S ×
(
¸
(

¸


=


The fission rate in the subcritical reactor is proportional to the neutron population,
so we can write these equations in terms of neutron power rather than neutron
population. Thus, the observed power level in the reactor is

source obs
P
k 1
1
P ×
(
¸
(

¸


=
where P
source
is the power that would be generated by the source neutrons
themselves in the absence of any multiplication by the fission process (that is, if ν
were equal to zero). P
obs
is the measured (actual) power level.
Variation of Source Multiplication
The factor [1/(1 – k)] is called the subcritical multiplication factor. Notice its
effect on a source. Even in a reactor that is well below critical, say ∆k = - 100 mk,
(k = 0.9) the equilibrium observed power level (P
observed
) is 10 times greater than
the actual photoneutron source power:
source source obs
P 10 P
9 . 0 1
1
P = ⋅
(
¸
(

¸


=
In this example, the subcritical multiplication factor is 10. This means that fission
produces the great majority of neutrons in the system and not the source directly.
The amount of subcritical multiplication depends only on the value of k. For
example, if the control absorbers fell into the core and the zones filled, the
reactivity insertion might be about -20 mk, giving a subcritical multiplication
factor of 50.
source source obs
P 50 P
980 . 0 1
1
P = ⋅
(
¸
(

¸


=
Notes:
CNSC 50
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

How Neutron Amplification, [1/(1-k)]
Changes as k approaches 1
1
10
100
1000
10000
0.400 0.500 0.600 0.700 0.800 0.900 1.000
neutron multiplication k
n
e
u
t
r
o
n

a
m
p
l
i
f
i
c
a
t
i
o
n

[
1
/
(
1
-
k
)
]

Figure 3.1
The Subcritical Multiplication Factor
Figure 3.1 shows the variation of the subcritical multiplication factor as k changes.
When a source is present, most of the neutrons in the reactor (at least for k > 0.5)
are not neutrons from the source, but neutrons that originate in fissions generated
because there are source neutrons. A glance at the figure shows that adding 100
mk to a reactor with k = 0.400 has less effect than adding, say, 20 mk to a reactor
with k = 0.950.
3.2 DYNAMICS IN THE SUBCRITICAL CORE
A positive reactivity insertion into a subcritical reactor increases power to a new
equilibrium level. The size of the increase and the time it takes for the power to
stabilize at the new value depends on the reactivity inserted and on the final value
of k. The closer k is to one, the larger the increase in power for a given reactivity
insertion, and the longer the time required to stabilize.
Figure 3.2 illustrates the increasing size and stabilization time for step reactivity
insertions that bring the core closer and closer to critical. The power level starts at
1 × 10
-5
% with k = 0.900. (The core is 100 mk subcritical). The figure shows nine
step reactivity insertions of +10 mk each, with P given time to stabilize after each
reactivity addition. (The numbers chosen are for illustration only; making 10 mk
steps would be quite impossible in practice.)
Notes:
CNSC 51
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003



Figure 3.2
Stabilization Time After Change in Multiplication Factor
Example:
The following example calculates the power changes in Figure 3.2 for two specific
+10 mk reactivity step increases as k approaches 1:
k changes from 0.900 to 0.910:
Power at k = 0.90 is
source
source
90 . 0
P 10
10 . 0
P
P = =
Power at k = 0.91 is
source
source
91 . 0
P 1 . 11
09 . 0
P
P = =
So, P
0.91
/P
0.90
= 1.11; power increases by 11%
k changes from 0.980 to 0.990:
Power at k = 0.98 is
source
source
98 . 0
P 50
02 . 0
P
P = =
Power at k = 0.99 is
source
source
99 . 0
P 100
01 . 0
P
P = =
So P
0.99
/P
0.98
= 2; power doubles
It is clear from this example that the change in power for a given reactivity
addition increases as k tends towards one.
Notes:
CNSC 52
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The fact that the stabilization time gets longer as k tends to one is more
subtle.
During normal subcritical reactor operation, moderator poisons keep the core
subcritical. The rate of reactivity addition by poison removal is slow, so we have
no need of “two group” kinematics equations. Instead, we will discuss time
dependence in the subcritical core using the average neutron generation-time
model. While the calculation is only approximate, it shows why the stabilization
time gets longer as the subcritical core gets closer to critical.
As shown earlier, the sum of an infinite series gives the total, stable neutron
population arising from a source. The series is
k 1
1
k k k k k k 1
2 n 1 n n 1 n 2

= + + + + + + + +
+ + −
L L
The sum exists because k is less than one, and each successive term is smaller than
the previous one. For practical purposes, the series ends after n generations if the
value of k
n
has become very small compared to the first term (that is, 1).
More precisely, if we pick a value of n that makes k
n
= 0.01, then the sum of the
entire series comes 99% from the terms before the n
th
term, and 1% from k
n

onward. [Writing the series as ) k k 1 ( k k k k 1
2 n 1 n 2
L L + + + + + + + +

, we see
that the series from the n
th
term onwards is just 0.01 of the entire sum].
We can calculate the number of generations, n, for the power to stabilize following
a change in k (that is reach within 1% of its equilibrium value) from the equation:
k
n
= 0.01
where k is the new value of k after a step insertion of reactivity.
n log k = log (0.01) = -2, or
( )
( ) ( ) k log
2
k log
01 . 0 log
n

= =
(Note: for convenience, use log to base 10)





Notes:
CNSC 53
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Table 3.2 shows the number of generations to stabilize for four selected
values of k. To determine approximate stabilization times, we take the time for
one generation to be the average neutron generation time (≈0.07s).

k # of generations
n = -2/log(k)
stabilization
time
0.800 21 1 s
0.900 44 3 s
0.990 458 30 s
0.999 4603 5 min
Table 3.2
Time to Stabilize
Clearly, as k→1, more neutron generations pass before the power stabilizes, and
the stabilization times get longer.
Formulas exist that will calculate the response of a subcritical reactor to insertions
of reactivity more accurately. Compare Figures 3.3 and 3.4. Each figure shows an
increase in subcritical power of about 25%, that is, about 0.1 decade (10
0.1

1.25).
The 60 mk step increase in reactivity in a deeply subcritical core produces a power
change in much less time than any reactivity device could, in practice, insert the
reactivity. In Figure 3.4, the power increase takes about five minutes for a small
reactivity insertion (+ 0.3 mk, making k = 0.9988) in an almost critical reactor. In
passing, notice that the more accurate analysis plotted in Figure 3.4 shows a
prompt jump, followed by a gradual rise. Notice also that the stabilization time is
comparable to the estimate in Table 3.2.
Notes:
CNSC 54
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

A +60 mk step into a Deeply Sub-Critical Core
(k = 0.400 ) - The Increase is 0.1 decades
1.00
1.10
1.20
1.30
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
time (seconds)
P
/
P
o

Figure 3.3
A Large Reactivity Insertion in a Deeply Subcritical Core
5% Zone Level Drop (+0.3 mk): Initially 1.5 mk Subcritical
1.00
1.10
1.20
0 60 120 180 240 300
time (seconds)
P
/
P
o

Figure 3.4
A Small Reactivity Insertion in a Nearly Critical Core
As the reactor approaches criticality, each successive reactivity insertion produces
a larger power increase, and the time to stabilize gets longer. This produces a
smooth transition between the subcritical and critical states. When deeply
subcritical the core is very sluggish; even large reactivity increases could go
unnoticed. An almost critical reactor behaves a lot like a critical reactor. Figures
3.5 (opposite) illustrate this.
The initial one minute of power rise is very similar in the two graphs of Figure 3.5.
However, power continues to rise in the critical core. The power reaches an
equilibrium power and stops rising in the subcritical core, where there is a single
power level for each -∆k, given by P
obs
= [1/(1-k)] P
source
Notes:
CNSC 55
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


An 0.2 mk step into an Almost Critical Core
(k = 0.9994 ) - Subcritical by about 8% zone level
1.00
1.10
1.20
0 15 30 45 60
time (seconds)
P
/
P
o

An 0.2 mk step into a Critical Core
(k = 1 )
1.00
1.10
1.20
0 15 30 45 60
time (seconds)
P
/
P
o

Figure 3.5
Comparison of a Critical Core with One that is Almost Critical
3.3 EXAMPLES
Determining Core Reactivity
We can use the subcritical multiplication equation to calculate the reactivity of the
core by observing the power before and after a known reactivity change. Usually
we do not know the source strength, but we do know it is constant on the short
term. We can write the equation for each observed power level and eliminate the
unknown source strength.
Notes:
CNSC 56
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Conversely, once you determine k, you can find the reactivity worth of a
device by observing the power level before and after inserting the device.
Suppose a reactor is shut down with a constant indicated power of 2 × 10
-5
%. The
operator inserts +1 mk by withdrawing an adjuster and power stabilizes at 3 ×
10
-5
%. Find the original value of k.
) factor tion multiplica initial k (
P
k 1
1
P
i
source
i
obs
=
(
¸
(

¸


=

Before the reactivity addition:
source
i
5
P
k 1
1
% 10 2
(
¸
(

¸


= ×


After the reactivity addition
( )
source
i
5
P
001 . 0 k 1
1
% 10 3
(
¸
(

¸

+ −
= ×


The source strength is the same in both cases:
P
source
= (1 - k
i
) × 2 × 10
-5
%, and
P
source
= [1 - (k
i
+ 0.001)] × 3 × 10
-5
%
Hence, equating to eliminate P
source

2(1 - k
i
) = 3 (0.999 - k
i
)
2 - 2 k
i
= 2.997 - 3 k
i

k
i
= 0.997
(Now, if you want to, you can substitute back and find out that the subcritical
multiplication factor initially was 333, and the source strength is 6 × 10
-8
%.)
Power Doubling
One example with important practical applications in reactor operation is the
power-doubling rule, stated as follows:
When a certain reactivity addition causes power to double in a subcritical reactor,
a further addition of the same reactivity would make the reactor critical.
To demonstrate why this is so, suppose we have a reactor that is subcritical with
an effective multiplication factor of k
i
(i = initial).

Notes:
CNSC 57
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The power level (P
i
) in the reactor is obtained by multiplying the source
power by the subcritical multiplication factor, and is
i
source
source
i
i
k
P
P
k 1
1
P
∆ −
= ⋅
(
¸
(

¸


=
where (k
i
– 1) = ∆k
i
is the amount by which the multiplication factor falls short of
criticality.
Now suppose we insert positive reactivity of ½ ∆k
i
so that the final reactivity is
∆k
f
= ½ ∆k
i
. Figure 3.6 shows the reactivities. The equation for the final power
level, P
f
, is
i
i
source
i
2
1
source
f
source
f
P 2
k
P
2
k
P
k
P
P =
∆ −
⋅ =
∆ −
=
∆ −
=
k -1
k
f

k
i

∆k
i
∆k
f


Figure 3.6
Change in Reactivity for a Power Doubling
Removing half the negative reactivity causes the power to double. Said another
way, when we observe a power doubling in a subcritical reactor, the amount by
which the reactor was subcritical has been cut in half.







Approaching Criticality by Power Doubling
Notes:
CNSC 58
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Table 3.3 demonstrates how, in principle, a series of power doublings
makes a deeply subcritical reactor approach criticality. Consider the first entry in
the table. The reactor is subcritical and initially k = 0.680 (∆k = -320 mk).
Measured power is 1.25 × 10
-6
of full power (10
-5.9
decades).

-∆k
(mk)
power
% F.P.
power
(decades)
0 320 1.25 × 10
-4
10
-3.9

1
st
160 2.5 × 10
-4
10
-3.6

2
nd
80 5 × 10
-4
10
-3.3

3
rd
40 1 × 10
-3
10
-3.0

4
th
20 2 × 10
-3
10
-2.7

5
th
10 4 × 10
-3
10
-2.4

6
th
5 8 × 10
-3
10
-2.1

7
th
2.5 1.6 × 10
-2
10
-1.8

8
th
1.25 3.2 × 10
-2
10
-1.5

9
th
0.6125 6.4 × 10
-2
10
-1.2
10
th
0.31 0.128 10
-0.9

Table 3.3
A Series of Power Doublings
Ten successive doublings raise the power by three orders of magnitude (2
10
= 1024
≈10
3
). Each doubling leaves half the negative reactivity in the core, so the reactor
is not yet critical. It is, however, within a few tenths of a mk from critical (the
equivalent of about 5% zone level). The regulating system could respond to a
request for a power increase and manoeuvre power at the demanded rate.
Each time power doubles in the subcritical core, noting the change in reactivity
that caused the doubling is, in effect, a measurement of k.
Figure 3.7 illustrates the principle of the power doubling. The graph assumes
startup instruments monitor count-rate (proportional to neutron power). The graph
is a straight line, because the subcritical multiplication formula
k
1
P
obs

∝ implies k
CR
1
∆ ∝
(
¸
(

¸

The point where the extrapolated line crosses the axis represents criticality. (when
∆k = 0, 1/CR = 0). The graph demonstrates visually that taking the reactor to
criticality by power doubling (i.e., by cutting [1/CR] in half), is a cautious way of
approaching criticality.

Notes:
CNSC 59
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

As long as one doubles power, the reactor gets closer and closer to critical
without actually going critical. By plotting the graph, or by keeping track of the
reactivity changes that double power, the operator can predict the reactivity worth
of the controlling device at criticality.
1/C
50,000
25,000
12,500
6,250
3,125
10 5 2.5 1.25
Negative reactivity (mk)

Figure 3.7
Approach to Critical by Power Doubling
Finally, Section 7 examines a commonly used criticality test that applies the
principle of power doubling. When the reactor is at low power it is difficult to tell
if the regulating system is holding power constant with k ≈ 1, or if the reactor is
subcritical with an observed power level corresponding to a particular -∆k.
The operator applies the test, typically once per shift at low power, asking the
regulating system to double power. If the liquid zones change excessively in
attempting to double power, the reactor is definitely subcritical. If a small
movement of zone level produces a power doubling, the operator knows the
regulating system could manoeuvre power to any demanded level at the demanded
rate, and is therefore, in effect, “critical”.
Notes:
CNSC 61
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

EFFECTS OF TEMPERATURE AND VOIDING
ON CORE REACTIVITY
4.0 INTRODUCTION
A change in power, for a reactor operating at high power, generally alters the
temperatures of the fuel, moderator, and coolant. A change in temperature of any
of these components causes a change in reactivity that, in turn, affects reactor
operation (a feedback effect). The temperature coefficients of reactivity (the
change in reactivity per unit change in temperature) determine the size and
direction of the reactivity change.
This section reviews the principal physical mechanisms that give rise to the
temperature coefficients. These are:
• Doppler broadening of the U-238 resonances;
• Changes in the thermal neutron energy spectrum;
• Density changes of the coolant and moderator.
Changes in each factor in the six-factor formula for k may contribute to the
temperature coefficient of a given reactor component (for instance, fuel). We
describe how the physical mechanisms above determine the sign of each relevant
contribution to both fuel and moderator temperature coefficients.
This section also looks at some operational implications of temperature
coefficients, including the changes of reactivity that occur on startup and
shutdown, and the importance of the Doppler Effect in providing negative
feedback to simplify normal regulation and to limit positive power transients.
Finally, the effects of void formation are an important source of feedback in power
transients. Voiding of the heavy water coolant from the fuel channels produces
positive reactivity. The practical implications of this include the need for a high
coolant isotopic.
4.1 FEEDBACK—AN INTRODUCTION TO TEMPERATURE EFFECTS E

At low power, when the fission rate is too low to generate much heat, addition of
positive reactivity produces an exponential rise, (that is, a constant log rate
increase). At power levels where there is significant heating, temperature changes
in the fuel, moderator, and coolant change reactivity—an effect that occurs quite
quickly. The following description demonstrates this reactivity feedback effect.
In 1949, a controlled experiment at AECL, Chalk River, allowed the NRX reactor
to “run away”. NRX was a heavy water moderated reactor with control rods for
reactor regulation. The heavy water level was set somewhat above the height at
which the reactor would be critical at low power with the rods withdrawn. The
Notes:
CNSC 62
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

experiment allowed reactor power in the supercritical reactor to increase
“unchecked”.
The manner in which the power changed was not what you might expect from
Section 2. At low power, the power did increase exponentially with a rate log of
3% P.P s
-1
(τ ≈ 33 s, ∆k ≈ +1.6mk). However, as the temperature of the fuel
increased, the reactivity decreased and caused the rate of power increase to slow
(see Figure 4.1). Later, the reactivity decreased even more, as the heavy water
became warmer. The total decrease in reactivity was enough, in this case, to make
the reactor subcritical, and so power reached a maximum and began dropping.

Figure 4.1
Self-Regulation Characteristics of the NRX Reactor
The reactor behaviour was self-regulating since temperature changes induced by a
power increase reduced reactivity and so prevented power from increasing
indefinitely. Of course, the initial excess reactivity in this experiment was quite
small; if it had been larger, the power likely would have continued rising.
The NRX experiment illustrates the effect of a negative temperature coefficient.
By definition, the temperature coefficient of reactivity is the change in reactivity
per unit increase in temperature. Its units are mk/°C or µk/°C (1 µk = 10
-3
mk). A
temperature coefficient may be positive or negative, but in the example, a
temperature increase caused a loss of reactivity so the coefficient was negative.
Temperature changes occur in the fuel, coolant, and moderator more or less
independently so there is a temperature coefficient of reactivity associated with
each.
It is very desirable for the overall temperature coefficient of a reactor to be
negative to provide the self-regulating feature illustrated by NRX. It is
particularly helpful if the fuel temperature coefficient is negative because in a
transient the fuel heats up more rapidly than the other core components do.
The reactivity effect of a temperature change is the product of the temperature
coefficient and the temperature change. It is helpful to know typical temperatures
of the fuel moderator and coolant in different operating states.
Notes:
CNSC 63
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Table 4.1 shows the effective average temperatures of fuel, coolant, and
moderator assumed in these notes. Temperatures in particular stations may differ
somewhat from these values.

Component Cold
Shutdown
Hot
Shutdown
Full
Power
Fuel 25 290 790
Coolant 25 265 290
Moderator 25 66 69
Table 4.1
Typical Temperatures of Reactor Components (°C)
As stated, the temperature coefficient is the change in reactivity per unit (°C)
change in temperature. The overall temperature coefficient is the slope of a graph
of reactivity (∆k/k) vs. temperature at the normal operating point. We will
introduce such graphs later in the section for moderator, coolant, and fuel.
Introducing
( )
dT
k d ln
= for the slope, where k = ηfpε Λ
f
Λ
t
, allows us to write
the temperature coefficient as a sum:
dT
d 1 ε
ε
+
dT
dp
p
1
+
dT
df
f
1
+
dT
d 1 η
η
+
dT
d
1
f
f
Λ
Λ
+
dT
d
1
t
t
Λ
Λ

To see how the physical effects of a temperature change influence the temperature
coefficient we need to recall the neutron life histories in the reactor and determine
the effect of temperature change on factors in the six-factor formula.
4.2 THE PHYSICAL BASIS FOR TEMPERATURE COEFFICIENTS
Later in this section, we look at the individual temperature coefficients of the fuel,
moderator, and coolant, and describe the effects of each temperature change on
each relevant factor in the six-factor formula. Before we do this, it is useful to
consider generally how temperature affects these factors. Three physical
processes produce the observed effects.
A temperature increase in a substance increases atomic and molecular motion. It
also (usually) causes thermal expansion. Each of these effects influences the
factors in the equation for reaction rate (R).
R = φΣ = Nσφ
where σ refers to the fission, absorption or scattering cross-section, as
appropriate.
Notes:
CNSC 64
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The three most important effects, in order of increasing importance, are:
• Thermal expansion decreases density and this directly affects N, the
number density of the nuclides present in the reactor. A subsection
below explains the effect in detail.
• Higher temperatures in the neutron’s environment change the
thermal neutron spectrum. This affects the thermal neutron cross-
sections for various scattering, absorption, and fission processes. A
subsection below explains these effects in detail.
• Increased molecular motion of U-238 increases the resonance
capture of neutrons. The motion increases the apparent width of the
resonances in the microscopic cross section, an effect known as
Doppler Broadening. A subsection below explains Doppler
Broadening and its effects.
Density Changes
Thermal expansion of the moderator as moderator temperature rises, decreases
moderator density. Since the number of atoms per unit volume (N) decreases,
neutrons travel further between interactions, so they diffuse over a wider range,
both while they thermalize, and as thermal neutrons.
A fuel density change does not have a large effect. Fuel bundles accommodate the
expansion of the fuel pellets so a change in fuel density may cause the reaction
rate per cubic centimetre of fuel to decrease, but the rate over the whole core does
not. By contrast, thermal expansion of heavy water displaces it from the core so
the amount of heavy water between the fuel channels decreases.
The effect of longer path lengths in the moderator and coolant, leading to
increased overall range of the neutrons, is an increased chance that neutrons will
leak out of the reactor. The fast and thermal non-leakage probabilities decrease,
tending to lower the reactivity, but the effect is not big.
A more important effect of longer path lengths in the moderator is that neutrons
are more likely to reach a neighbouring channel before the moderator has
thermalized them. Resonance capture increases so resonance escape probability
decreases.
On the other hand, the reduction in atomic density lowers the macroscopic
absorption cross-sections in the moderator or coolant, which increases the thermal
utilization (f), thus tending to increase reactivity. The effect on f of the reduction
in atomic density is particularly strong in the moderator if it contains an
appreciable quantity of poison.

The Neutron Energy Spectrum
Notes:
CNSC 65
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

We usually assume all thermal neutrons have energy of 0.025 eV and
travel at 2200 m/s. In fact, thermal neutrons, by elastic collisions, distribute
themselves over a range of energies. On average, the neutrons share the kinetic
energy of the atoms and molecules with which they interact, but any particular
elastic collision can increase or decrease the neutron kinetic energy. A rather
complicated mathematical expression known as the thermal neutron spectrum
shows the variation in neutron density as a function of thermal neutron energy.
We associate a neutron temperature with each such distribution; the temperature
depends on the temperature of the surroundings.
A temperature change of any component of the reactor affects the neutron
spectrum. This, in turn, changes fission and absorption rates in the fuel and
absorption rates throughout the reactor, because cross sections vary with neutron
energy. The following subsections provide a step-by-step guide to the effects of
spectrum changes on reactivity.
The Thermal Neutron Spectrum
Figure 4.2 shows thermal neutron spectra for 20°C and 300°C. Thermal neutrons
in a CANDU core will always have spectra somewhere between the distributions
shown. (At full power, the thermal neutron temperature of neutrons interacting
with CANDU fuel is about 200 °C. When shut down, the neutron temperature will
be hotter than 20°C.) Figure 4.2 is a Maxwellian distribution of neutron speeds,
plotted as a function of neutron energy, with a logarithmic scale that allows easy
comparison with cross-section graphs.
0.010
0.100
1.000
0.0001 0.0010 0.0100 0.1000 1.0000
Neutron Energy (eV)
#

o
f

n
e
u
t
r
o
n
s

(
a
r
b
i
t
r
a
r
y

l
o
g

s
c
a
l
e
)
300°C
20°C

Figure 4.2
Neutron Energy Shift with Changing Temperature

Notes:
CNSC 66
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

At room temperature, the most probable value for thermal neutron energy
(the peak of this curve) is 0.0253 eV, corresponding to a speed of 2200 m/s. The
average neutron energy is proportional to the absolute temperature, so the shift in
the spectrum for a temperature change from 20°C to 300°C (293 K to 573 K)
almost doubles the energy at the peak in Figure 4.2. The shift of the neutron
spectrum to higher temperatures chiefly affects the reproduction factor (η), as
discussed below, because the main effect of changing the neutron spectrum is to
change the rates of thermal neutron absorption in the fissile isotopes.
Reaction Rates in Fissile Isotopes
Many of the materials in the reactor have absorption cross-sections that vary
inversely with the neutron speed. As neutron speeds increase, the number of
thermal neutrons passing through a region each second increases (φ = nv) but the
increased neutron speed reduces the target size (σ
a
∝ 1/v). This means that the
absorption rate, R
a
= Nσ
a
φ, does not change for 1/v cross-sections.
If all materials exhibited the same 1/v dependence on neutron speed, changes in
neutron temperature would not affect reaction rates and there would be no
significant reactivity effect to discuss. However, the absorption cross-sections of
the fissile isotopes deviate significantly from l/v behaviour.
Figure 4.3 illustrates this with a plot the absorption cross-sections of U-235 and
Pu-239 relative to that of U-238. The U-238 cross-section is very nearly l/v, so the
rate of absorption in U-238, as for all 1/v absorbers, hardly changes as the
temperature of the thermal neutrons change. The ratios plotted show how the
absorption rates in U-235 and Pu-239 change with temperature.
Notes:
CNSC 67
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

0.90
1.00
1.10
1.20
1.30
1.40
1.50
0 50 100 150 200 250 300
Temperature
o
C
C
r
o
s
s

S
e
c
t
i
o
n

R
a
t
i
o
Pu239
σa
239
/σa
238
U235
σa
235
/σa
238
Cross Section Ratios are Normalized to 1
at thermal neutron energy, 20
o
C,
for easy comparisons

Figure 4.3
Deviation of Absorption Cross-Sections of Pu-239 and U-235 from 1/v
Figure 4.3 shows that, by comparing to U-238, the thermal neutron absorption rate
in U-235 drops a little as the neutron temperature increases. Looking at R = Nσφ,
the absorption cross-section for U-235 drops off a little faster than 1/v, while the
increase in thermal flux is proportional to v (φ = nv).
For Pu-239, on the other hand, an increase in neutron temperature causes a very
strong increase in the absorption rate in Pu-239. Although the absorption cross
section is dropping (see Figure 4.4), it is not decreasing nearly as much as 1/v.

Figure 4.4
Pu-239 Fission Cross-Section


Notes:
CNSC 68
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

For Pu-239, the absorption rate in a CANDU increases with neutron
temperature mainly because the cross-section does not drop enough in the thermal
energy range. (The bulk of the thermal neutrons are in the energy range below
about 0.1 eV.) The large resonance centred at 0.3 eV enhances the absorption rate,
because a few percent of the neutrons at the high end of the thermal spectrum
move into the resonance region as the temperature rises.
In summary, as the neutrons get hotter, the rate of absorption, both fission and
capture increases strongly in Pu-239, and they drop a little in U-235.
Variation of η with neutron energy
Unfortunately, knowing how the absorption rates for the fissile isotopes change
with neutron temperature does not tell us, directly, how η (a ratio of cross-
sections) changes for equilibrium CANDU fuel, which is a mixture of isotopes.
The reproduction factor for CANDU fuel is the ratio of an effective flux weighted
average over all the fuel contents:
( )
( ) fuel
fuel
a
f
Σ
Σ
ν = η
The cross sections for both U-235 and Pu-239 affect the numerator and
denominator. Each changes differently with neutron temperature, so we need to
re-write this expression to show the contributions of each fissile nuclide. It turns
out that reproduction factor for CANDU fuel is a weighted average of the
reproduction factors for U-235 and Pu-239, as follows.
) fuel (
a
9
a 9
5
a 5
Σ
Σ η + Σ η
= η
where the labels 5 and 9 refer to U-235 and Pu-239 respectively. (We’ve ignored
Pu-241).
The reproduction factors η
5
and η
9
are for the pure isotopes—ν×(σ
f

a
). The
temperature dependence of the microscopic cross-sections and the reproduction
factors for the pure isotopes is well known and many textbooks give them. Figure
4.5 shows how η
5
and η
9
change as temperature changes. The upper line in the
figure is a horizontal line, to make clear that η
5
decreases a little with temperature.
The reproduction factor for Pu-239 drops even more than does η
5
, but even for
Pu-239, η decreases only about 7% between 20°C and 300°C.
Notes:
CNSC 69
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

1.90
1.95
2.00
2.05
2.10
0 50 100 150 200 250 300
Temperature
o
C
η
(
T
)
U235
Pu239

Figure 4.5
Reproduction factors of U-235 (η
5
) and Pu-239 (η
9
)
We can now assemble the pieces to see how η for CANDU fuel varies as the
neutron temperature increases. For fresh fuel bundles with only U-235 in the fuel,
the situation is straightforward. As the thermal neutron temperature increases,
there is a decrease in both the neutron absorption rate in U-235 (Figure 4.4) and
the number of neutrons returned per absorption in U-235 (η
5
, Figure 4.5). Neither
of these variations is very pronounced. It is clear that the contribution of U-235 to
the temperature coefficient is negative, but it is not very big.
For fuel with some Pu-239 present, an increase in neutron temperature greatly
increases the absorption rate in Pu-239. This outweighs the small decrease in η
9

that occurs at the same time. The contribution of Pu-239 to the overall
temperature coefficient is strongly positive. The effect is so large that for an
equilibrium-fuelled reactor, it easily dominates the negative U-235 contribution,
so:
dT

η
1
is positive for equilibrium CANDU fuel.
Other Spectrum Effects
The only factors other than η that the thermal neutron spectrum might affect
directly are the thermal utilization factor (f) and thermal non-leakage probability
factor Λ
t
. (Other factors in the six-factor formula describe fast and intermediate
energy neutrons.)
Section 1 gives the thermal utilization factor:
( )
( ) ( ) fuel non fuel
fuel
f
a a
a
⋅ Σ + Σ
Σ
=

Notes:
CNSC 70
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

As we have seen, the Pu-239 content of equilibrium fuel causes a strong
increase in neutron absorption as the neutron temperature increases. The non-fuel
absorption is almost independent of temperature, so the contribution to reactivity
of thermal utilization is positive for equilibrium CANDU fuel. The size of this
reactivity increase is not nearly as big as the contribution from η.
Thermal neutron leakage increases a little for warmer thermal neutrons. All
absorption cross sections drop as neutron energy increases (whether or not the
decrease varies as 1/v). Consequently, the mean free path to absorption for
thermal neutrons increases (similar to the effect of a density decrease, but
restricted to the thermal neutrons). They wander a bit farther from their starting
points and are a bit more likely to leak out.
Doppler Broadening in the Fuel
The Doppler Effect arises directly from a temperature change in the fuel.
Resonance capture in U-238 increases as the fuel gets hotter for the following
reason. The absorption cross-section of U-238 in the resonance region consists of
a set of sharp peaks of the sort shown in Figure 4.13. In the resonance region, the
probability that a stationary U-238 nucleus absorbs a neutron depends on whether
or not the neutron kinetic energy is at the exact energy of a resonance. Both the
nucleus and neutron are moving, so the crucial variable in determining the
probability of absorption is the speed of the neutron relative to the U-238 nucleus.
Heating the fuel makes the atoms of U-238 move more vigorously, so the relative
speeds of the neutrons and the U-238 nuclei change.
To demonstrate the effect, consider the stationary nucleus shown at the left of
Figure 4.6. A neutron with a speed corresponding to the peak of the resonance has
a high probability of being absorbed, while the stationary nucleus is unlikely to
capture a neutron travelling slightly slower or faster than this.
Now think about what happens when the fuel is heated and the U-238 nuclei are
moving vigorously. A neutron whose speed is such that it previously lay well
outside the peak may encounter a U-238 nucleus that is moving at that instant in
such a way that the speed of the neutron relative to the nucleus coincides with the
peak.
Notes:
CNSC 71
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003




Figure 4.6
Mechanism of Doppler Broadening
Here E
n
represents the kinetic energy of the neutron and E
res
the energy at the peak
of the resonance. All three neutrons appear to the nucleus as having the resonance
energy.
The centre diagram in Figure 4.6 shows a neutron with speed higher than in the
left-hand diagram. However, the neutron encounters a nucleus moving away at a
speed that makes the neutron’s speed of approach the same as in the first diagram.
To the nucleus, this neutron appears to be at the resonance peak energy. The same
thing can happen to a neutron with speed below the peak of a nucleus at rest, but
which happens to encounter a “hot” uranium nucleus moving towards it at exactly
the right speed (right-hand diagram).
Heating the fuel effectively “broadens” the resonances as shown in Figure 4.7. At
the same time, increased motion of the U-238 atoms reduces the height of the peak
because neutrons moving at exactly the correct speed originally are no longer
moving at that speed relative to the moving nuclei. The overall absorption might
not change very much if the reduced rate of absorption at the peak were to offset
the increased absorption away from the peak.









Notes:
CNSC 72
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

However, the U-238 cross-section is so high in any case that a neutron
within the overall peak region is almost certain to be absorbed (although it will
likely travel further into the fuel first.)

Figure 4.7
Doppler Broadening of U-238 Resonance
Put another way, the peaks are sharper when the fuel is not as hot. With a sharp
peak, neutrons at the resonance energy are absorbed close to the fuel surface, (mfp
≈ 0.05 mm in UO
2
) and neutrons away from the peak energy are likely to pass
through the fuel to continue slowing in the moderator. (The surface absorption
self-screens the bulk of the fuel from resonance energy neutrons). Hotter fuel
exposes a large volume of U-238 to neutrons with energies slightly different from
resonance peak, and many of these are absorbed in addition to the absorption of
the resonant energy neutrons.
The net effect of heating, however one describes it, is to broaden the range of
neutron energies (i.e., the numbers of neutrons) with a high probability of
resonance capture. The reduced resonance escape probability (p) results in
reduced reactivity.
4.3 TEMPERATURE COEFFICIENTS OF REACTIVITY
We have now introduced the physical effects needed to discuss each of the
reactivity coefficients in turn.
The Moderator Temperature Coefficient of Reactivity
Figure 4.8 shows the change in core reactivity as moderator temperature changes.
(The curves shown embrace a range of possible equilibrium fuel conditions.
Section 6 introduces the burnup unit, n/kb.) The reactivity change in going from
25°C to an operating temperature of 75°C is about 4 mk, for an average
temperature coefficient of 4 mk/50°C = +80 µk/°C.

Notes:
CNSC 73
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The slope near the normal operating temperature is a little less than this
average value, about 75 µk/°C.


Figure 4.8
Change in Reactivity with Moderator Temperature
One practical effect of the positive moderator temperature coefficient is the
possibility of increasing reactivity by adjusting the moderator temperature set
point, so that the Moderator Temperature Control program drives the temperature
up. For example, suppose the moderator temperature is normally 60°C and the
temperature is raised to 65°C. For a temperature coefficient of 75 µk/°C, this
produces a reactivity increase of 0.38 mk (5°C × 75 µk/°C = 0.38 mk). This
amount of positive reactivity is equivalent to a day or so of fuelling, and is a
possible response to unexpected fuelling machine unavailability.
There are few circumstances where it would be possible to gain more than this, as
the temperature of the moderator is limited, usually below about 70°C (as
measured at the outlet) for safety reasons. Safety Analysis credits the moderator
as a “last ditch” heat sink—heat removal by moderator water would limit fuel
failures in an accident where no other cooling is available. The analysis requires
an upper limit on moderator temperature so that cooling will be effective if it is
required.
The explanation of the increase in reactivity with moderator temperature is as
follows. Heating the moderator produces two important physical effects: (a) a
decrease in moderator density and, (b) an increase in the average neutron energy.


Notes:
CNSC 74
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The temperature of the moderator has a big effect on the neutron
temperature, so we might expect the reproduction factor (η) to dominate the
moderator temperature coefficient. We have seen that the reproduction factor
makes a positive contribution to the reactivity as neutron temperature increases.
When moderator density decreases, the resonance escape probability (p) decreases.
The increase in distance between moderator atoms increases the distance a neutron
travels while slowing down; so more neutrons reach neighbouring channels before
the moderator has completely thermalized them. (Longer neutron path lengths
also increase fast and thermal leakage, but the decrease in the non-leakage
probabilities is quite small).
Lower moderator density decreases thermal neutron absorption in the moderator.
This results in a positive reactivity contribution from f. The changes in f and p
tend to cancel each other but there is a net increase in reactivity that adds to the
increase in reactivity from η. The reason the increase in f outweighs the decrease
in p is because CANDU reactors are overmoderated. Adequate thermalization
occurs even with fewer moderator atoms between the fuel channels, so the main
effect of reducing the amount of moderator is a decreased chance of absorption.
The reduction in absorption in the moderator is particularly pronounced when
control poisons are present. In this case, there is a much larger reactivity change
with temperature than is shown in Figure 4.8.
Notes:
CNSC 75
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


The Coolant Temperature Coefficient of Reactivity
Figure 4.9 shows the effect of changing the coolant temperature. For equilibrium
fuel, the coefficient is positive throughout the whole temperature range. Its value
near the normal operating temperature (the slope of the curve at that point) is
typically about 40 µk/°C. The control system does not adjust coolant temperature
independently of the reactor power, but coolant temperature changes a little as the
fuel temperature changes.


Figure 4.9
Change in Reactivity with Coolant Temperature
From the point of view of the neutrons, the coolant is just more moderator. (The
pressure tubes and calandria tubes are nearly transparent to neutrons). The same
kinds of reactivity effects that occur in the moderator also occur in the coolant.
The sizes of the various contributions are different from the moderator
contributions because of the much smaller volume of coolant, the geometry
(coolant close to the fuel) and higher operating temperature.
The Fuel Temperature Coefficient of Reactivity
Figure 4.10 shows the effect of changing fuel temperature, as occurs on any power
manoeuvre. The average value of the fuel temperature coefficient between hot
shutdown and full power is about - 8 µk/°C. Near the normal operating
temperature, the slope of the graph is about - 4 µk/°C.


Notes:
CNSC 76
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The fuel temperature coefficient arises principally from two factors: one is
the Doppler Effect and the other is the neutron spectrum effect. The Doppler
Effect reduces the resonance escape probability and this effect on reactivity is very
much larger than all other reactivity effects in the fuel. The neutron spectrum
shifts to a higher temperature because of the hotter fuel, so for equilibrium fuel,
the positive η coefficient partly offset the strong negative Doppler contribution.


Figure 4.10
Change in Reactivity with Fuel Temperature.
Notes:
CNSC 77
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


4.4 REACTIVITY VARIATION WITH TEMPERATURE
Figure 4.11 shows the variations in overall reactivity in a CANDU 600 as the
power changes from cold shutdown to full power. (The figure gives curves for
both fresh and equilibrium fuel.). With equilibrium fuel, the net reactivity change
in going from cold shutdown to full power is less than +2 mk.

Figure 4.11
Reactivity Changes with Temperature of Reactor Components
The Power Coefficient
Of particular interest in Figure 4.11 is the reactivity loss of about 3 mk in going
from the hot shutdown condition to full power. The change in reactivity in
moving from the hot shutdown to the 100% power condition is known, somewhat
confusingly, as the power coefficient. This “coefficient” is the total reactivity
change between the two states rather than the reactivity change per °C, as was the
case for the coefficients discussed earlier. The CANDU power coefficient is
typically about -2mk to -4mk for equilibrium fuel.
To increase power from hot shutdown to full power, the regulation system, in
addition to providing positive reactivity to increase power, must increase reactivity
by +3 mk to offset the loss due to heating. The regulating system will insert much
smaller amounts at regular intervals to ramp the power up. As the power rises, the
loss of reactivity from heating tends to offsets the reactivity insertion. A reactivity
insertion big enough to start the power rise is soon overcome by the negative
power coefficient, so, for a moderate rate of power increase, the regulating system
is continuously playing “catch up” to keep the power rising at the demanded rate.
Notes:
CNSC 78
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

There is little risk of overshooting the target power, and the regulating
system achieves smooth control without complicated feedback circuits.
Conversely, small negative reactivity insertions do not shut the reactor down.
Suppose a rod falls accidentally into the core, inserting one mk of negative
reactivity. Ignoring the regulating system response to this, the loss of 1 mk “uses
up” about 1/3 of the power coefficient, so the power drop will slow and stop
somewhere below 70% full power, with the reactor critical.
Example
We can calculate the overall reactivity change in going from one condition to
another, given the temperature coefficients and temperature changes for each
component. Let us estimate the loss of reactivity in going from hot shutdown to
full power (the power coefficient). The values of the temperature coefficients
involved, averaged over the appropriate temperature ranges, are:
Fuel coefficient: -8 µk/°C
Coolant coefficient: +40 µk/°C
Moderator coefficient: +80 µk/°C
Using the temperature values in Table 4.1, we expect effective average fuel
temperature to rise about 500°C from hot shutdown to full power. The coolant
temperature may rise about 25°C, (except for the Pickering reactors, which ramp
boiler pressure down as power rises, to keep the average coolant temperature
nearly constant). Moderator temperature control keeps the moderator temperature
nearly constant. (Typically, moderator temperature control keeps the outlet
temperature constant, but the bulk temperature rises a little.) Any change in
moderator temperature that does occur is relatively slow because of the large mass
of moderator water.
The expected reactivity change in moving from hot shutdown to full power is the
sum of the contributions from Fuel, Moderator, and Coolant.
Fuel: -8 µk/°C × 500 °C = -4 mk
Coolant: +40 µk/°C × 25°C = +1.0 mk
Moderator: +80 µk/°C × 0°C = 0 mk
TOTAL -3 mk

Self Regulation
Notes:
CNSC 79
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

To see whether the fuel, moderator, and coolant temperature coefficients,
working together, provide the sort of “self regulation” described in Section 4.1, we
must consider:
• The size of the three temperature coefficients;
• The size of each temperature change when a power change occurs;
• The time for each component to heat up.
In a fast power transient, the fuel temperature increase is much larger and occurs
more immediately than any coolant temperature increase. The coolant temperature
is likely to lag the fuel temperature by several seconds. Therefore, the negative
fuel temperature coefficient provides some degree of the desired self-regulation.
This effect arises because of Doppler broadening of the U-238 resonances in the
fuel as the fuel heats.
Ignoring the response of the regulating and safety systems, after an upset that
inserts a small amount of positive reactivity into a critical core at low power,
power will not rise indefinitely. For a large reactivity upsets, the power may
continue to rise beyond full power, but the power coefficient reduces the rate of
rise, giving time for automatic safety system response.
4.5 VOID REACTIVITY
Voids form if the moderator or the heat-transport system fluid boils. Void
formation is more likely to occur in the pressurized coolant than in the moderator,
so we will restrict our discussion to the effects of loss of coolant.
Reactivity increases with loss of heavy water coolant, and this produces a power
pulse just as heat removal capability is deteriorating. The large loss of coolant
accident (large LOCA) defines many of the requirements of the emergency safety
systems.
The overall reactivity effect of voiding the coolant is the insertion of positive
reactivity. The total reactivity change for full core voiding is typically in the range
of +7 to +13 mk, depending on the model of CANDU. Notice that the size of this
reactivity is more than enough for prompt criticality.
Fortunately, it takes time for all the coolant to flash to steam through a rupture;
long enough for emergency shutdown instruments to detect the power increase and
trigger a reactor trip. High rate log, measured by shutdown system ion-chambers,
or reactor power high, measured by in-core detectors, are the most likely trips.


Possible causes of coolant boiling are:
Notes:
CNSC 80
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

• Low pressure (pipe rupture, pressurization system failure);
• Low flow (blockage, pipe rupture, pump failure);
• Excess power (flux distortion, regulating system failure).
Under these circumstances, steam gradually displaces the liquid coolant and
eventually the channel(s) become totally depleted of liquid.
Voiding fuel channels significantly affects:
• The fast fission factor (ε);
• The resonance escape probability (p);
• The reproduction factor (η); and
• The thermal utilization factor (f).
Of these, the first two are the main contributors to the positive void reactivity.
Voiding causes a decrease in the moderation of neutrons in the immediate
neighbourhood of the fuel elements. Looking at Figure 4.12, you can see that a
neutron born in one fuel element (for instance, element A) normally passes
through some coolant before reaching the next fuel element (element B); coolant
molecules usually begin the moderating process. The first elastic collision of a
fission neutron with a deuterium nucleus reduces the neutron energy to below the
fast fission threshold. A few collisions could reduce the energy into the resonance
energy range. (Figure 4.13 shows the radiative capture and fission cross-sections
of U-238.) Without coolant present, neutrons in the fuel channel are more likely
to be above 1.2 MeV and less likely to be below 10 keV than if coolant is present.
Notes:
CNSC 81
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Figure 4.12
Disposition of Moderator and Coolant around the Fuel Bundle

Figure 4.13
Capture and Fission Cross-Sections of U-238




Notes:
CNSC 82
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The removal of coolant produces the following effects:
• An increase in the fast fission factor (ε), since more fission
neutrons have a chance of interacting with U-238 while still above
the threshold energy for fission;
• An increase in the resonance-escape probability (p), since fewer
fission neutrons reach resonance energy before escaping from the
fuel channel.
Normally, thermal neutrons drifting back into the fuel channel from the moderator
often interact with coolant molecules before the fuel absorbs them. This “re-
warming” of the neutrons by the coolant hardens the spectrum. Voiding removes
hot coolant from the channel and this reduces the thermal neutron temperature.
Without coolant present, the thermal neutrons absorbed by the fuel are “cool”
neutrons from the moderator that have not been “re-warmed” by interactions with
the coolant. This thermal neutron spectrum change causes a change in η that is
not quite as large as the reactivity effect from ε or from p.
• In this case, since we are dealing with a decrease in neutron
temperature, the reproduction factor (η) decreases with voiding for
equilibrium fuel.
• Voiding the coolant decreases the amount of absorbing material in
the reactor (mainly light water impurity in the coolant), increasing the
thermal utilization (f). Keeping the coolant isotopic high limits the
increase in f on voiding.
License Limits on Coolant Isotopic
Downgrading the moderator water (increasing its light-water impurity) quickly
results in the reactor going subcritical because of the high neutron absorption.
Low coolant isotopic does not physically prevent reactor operation since relatively
fewer of the neutron collisions occur in the coolant. Consequently, administrative
procedures enforce a lower coolant isotopic limit, imposed by the operating
licence. The limit is usually in the range of 97.5%. If the isotopic is near this
license limit, the size of the contribution of f to the void reactivity is comparable to
the other contributions. If the isotopic decreases below the allowed limit, f can
quickly become the largest contributor to an unacceptably large, positive void
coefficient.
An excessively large positive void reactivity causes a large power surge during the
void formation. This is likely to cause severe damage to the reactor if the
protective systems cannot respond adequately. Trip set points, amount of negative
reactivity, and rate of insertion are safety shutdown system parameters designed to
accommodate the worst-case accidents; analysis of such accidents determines the
operating limits on parameters, such as coolant isotopic.

Notes:
CNSC 83
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This is a convenient place to mention the upper limit on coolant isotopic.
Accident analysis demonstrates the effectiveness of backup trip parameters for
each analysed accident. One of the analysed accidents is the rupture of a pressure
tube, causing calandria tube failure such that coolant (at high pressure) displaces
moderator water. If this occurs during startup with a heavy poison load in the
moderator, the displacement of neutron absorbing moderator water with clean
coolant causes a large reactivity increase. This should result in a trip on high rate
of power increase. It is also necessary to ensure a timely backup trip. By keeping
the coolant isotopic lower than that of the moderator (or setting an upper limit on
coolant isotopic), the rate of positive reactivity addition is decreased slightly in the
accident just described, allowing time for a process trip.
The process of upgrading moderator and coolant heavy water tends to favour
upgrading moderator water, simply because the saving in fuel cost is higher for
upgraded moderator water. Even in the days before there was an upper limit on
coolant isotopic, the moderator isotopic was usually higher than the coolant
isotopic. For coolant isotopic lower than moderator isotopic, the reactivity effect
of displacing the poisoned moderator, while still high, is not quite as high.
Accident analysis credits this slight reduction in reactivity in demonstrating that
the backup trips occur in a timely fashion, so the Operating Policies and Principles
include a requirement to keep coolant isotopic below moderator isotopic (or
specifies an upper limit on coolant isotopic).

Notes:
CNSC 85
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

EFFECTS OF FISSION PRODUCTS ON CORE
REACTIVITY
5.0 INTRODUCTION
All fission products absorb neutrons to some extent, so are known as reactor
poisons. Most fission product poisons simply build up slowly as the fuel burns up
and are accounted for as a long-term reactivity. The neutron absorbing fission
products xenon-135 and samarium-149 have particular operational importance.
Their concentrations can change quickly, producing major changes in neutron
absorption on a relatively short time scale. Each arises from the decay of a
precursor fission product, which controls their production rate, but, because they
have large absorption cross-sections, their removal changes quickly with changes
in flux.
We start by considering the mechanisms for creation and destruction of xenon and
its precursor, iodine-135. This allows us to derive expressions for the steady state
concentrations of I-135 and Xe-135. It also enables us to analyze the sequence of
events following a shutdown after prolonged operation at power, and to
understand why this leads to a rapid increase in Xe-135 concentration.
Any power manoeuvre produces a transient change in xenon concentration.
Reactivity changes caused by xenon concentration changes are not as immediate
as the reactivity changes caused by a change in fuel temperature, but the size of
the effect can be much larger. Unlike temperature feedback, feedback from xenon
is positive. For example, a power increase causes an increase in reactivity that
causes a further power increase. This is important because the reactivity effect can
exceed the capability of the automatic control system to compensate for it.
Another important consequence of the presence of Xe-135 in a CANDU reactor is
that it leads to the possibility of xenon oscillations. These can cause reactor power
to rise and fall with a period of 15-30 hours, with the possibility of over-rating the
fuel. We will describe the process by which xenon oscillations can occur, and see
why they necessitate continuous flux monitoring at a number of points in the
reactor.
Finally, we will look at the effects of Sm-149. Although it does not create as
dramatic an effect as Xe-135, it contributes a significant reactivity at equilibrium,
and builds up to a considerably larger load after shutdown. Unlike the xenon
reactivity, this does not subsequently disappear, because Sm-149 is a stable
isotope.

5.1 XENON AND IODINE BUILDUP
Notes:
CNSC 86
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Xenon-135 (often simply referred to just as xenon) is the most important
fission product poison. It has a very large absorption cross-section and high
production rate.
Fission produces Xe-135 in two ways:
Directly from fission—about 0.6% of all fissions in equilibrium CANDU fuel
produce Xe-135.
Indirectly from the decay of I-135—fission produces I-135 directly, or as a fission
product daughter of tellurium-135. About 6.4% of all fissions in CANDU fuel
produce either I-135 or Te-135.
The entire decay chain is as follows:
γ + β + →

I Te
135
53
135
52
t
½
= 19 s
γ + β + →

Xe I
135
54
135
53
t
½
= 6.6 h
γ + β + →

Cs Xe
135
55
135
54
t
½
= 9.1 h
Due to the short half-life of Te-135, we normally consider the entire 6.4% fission
product yield to be I-135.
The decay of I-135 to Xe-135 is, in practice, the only way in which iodine is lost.
(Removal of I-135 by radiative capture is negligible compared to its decay because
I-135 has a very small absorption cross section).
There are two loss mechanisms for xenon; both burnup and decay are important
removal processes. At high power, neutron-capture removes much more xenon
than does beta decay. An important point is that burnup changes immediately
when flux changes, while the 9.1-hour beta decay half-life governs its decay rate.
Xenon Reactivity Effect
Xenon is a strong neutron absorber so its presence in the fuel creates a large
negative reactivity in the core. The reactivity worth of the Xe-135 is known as the
xenon load. At full power, the reactivity worth of equilibrium xenon in a large
CANDU is about -28 mk. This reactivity changes by only a couple of mk above
50% of full power.
It is also common practice to express the concentration of iodine as iodine load (in
mk). It is important to realize that iodine alone is not a significant poison; there is
no appreciable reactivity associated with it.
The definition of iodine load is the reactivity that it would insert into the reactor if
all the iodine present suddenly changes into xenon. Note this point carefully and
Notes:
CNSC 87
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

understand that we are not talking about a real reactivity that exists in the
system, but a reserve bank of potential reactivity that will gradually, and
inevitably, become neutron-absorbing Xe-135. For a large CANDU, the
equilibrium iodine load at full power (the iodine inventory, or reserve “bank”) is
worth about -320 mk. The value for any particular reactor depends on the full
power flux.
Balancing Xenon and Iodine Production and Removal
We now look at the processes of production and loss that determine the buildup of
I-135 and Xe-135 when reactor power increases. You may find it useful to refer to
Figure 5.1, a block diagram that depicts the production and loss rates of the two
isotopes.

Figure 5.1
Production and Loss Mechanisms for I-135 and Xe-135
The rates of production depend on the fission rate per unit volume:
R
f
= φΣ
f

where φ is the average thermal neutron flux in the fuel and Σ
f
is the macroscopic
fission cross-section. Typical values of flux and macroscopic fission cross section
in the central region of a large CANDU at high power are φ ≈ 10
14
n cm
-2
s
-1
and
Σ
f
≈ 0.1 cm
-1
.

Iodine-135
Concentration

(NI )
Xenon-135 Concentration
(N
Xe
)
Iodine:
Direct Fission Production
(γIΣfφ )
Xenon:
Direct Fission Production
(γXeΣfφ )
Iodine Decay
(λINI)
Xenon Burnup
(NXeσaφ)
Xenon Decay
(λXe NXe)
Notes:
CNSC 88
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Table 5.1 shows the microscopic cross-sections, yields, and half-lives for
xenon and iodine.
The rate of production per unit volume of I-135 from fission R
f
(I-135) = γ
I
Σ
f
φ
where γ
I
is the fission product yield of I-135 (that is, the fraction of fissions that
gives rise to I-135).
Depletion of I-135 occurs by its decay to Xe-135. The decay rate = λ
I
N
I
, where λ
I

is the decay constant of I-135, and N
I
is its concentration in atoms per cm
3
.
Subtracting the I-135 loss rate from its production rate gives the net rate of change
of iodine concentration at any time. We therefore have (see the upper block in
Figure 5.1)
( )
I I f I I
N N
dt
d
λ − φ Σ γ =
For xenon-135, the equation is a little more complicated. We have two production
terms, since fission produces xenon directly and as a fission product daughter. We
also have two loss mechanisms; both burnup and decay are important removal
processes for xenon.
The rates of Xe-135 production (per unit volume) for each production method are:
Direct fission production
R
f
(Xe-135) = γ
xe
Σ
f
φ where γ
xe
is the Xe-135fission product yield.
Iodine decay = λ
I
N
I
.
The rates of removal of Xe-135 (per unit volume) for each loss mechanism are:
Burnup
R
a
(Xe-135) =
Xe
Xe
a
N φ σ
Xenon decay = λ
xe
N
xe

where N
Xe
is the concentration of Xe-135 in atoms per cm
3
, λ
Xe
is its decay
constant, and
Xe
a
σ is its microscopic absorption cross-section.




Notes:
CNSC 89
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Iodine-135 Xenon-135
σ
a
(barns) negligible 3.5 × 10
6

fission yield 6.4% 0.6%
t
½
(hours)
6.6
9.1
λ (s
-1
) 2.93 × 10
-5
2.1 × 10
-5

Table 5.1
Properties of Iodine and Xenon
Neither Xe-136, formed when Xe-135 captures a neutron, nor Cs-135, from
Xe-135 decay, is a strong neutron absorber, so xenon removal reduces neutron
absorption in the fuel.
Subtracting the Xe-135 loss rate from its production rate gives the net rate of
change of xenon concentration at any time. We therefore have (see the lower
block in Figure 5.1)
( ) | | | |
Xe Xe Xe
Xe
a I I f Xe Xe
N N N N
dt
d
λ + φ σ − λ + φ Σ γ =
Iodine and Xenon: Equilibrium Buildup
Looking at the I-135 equation, we see that if we start a reactor (raise the flux to a
high, constant value) with no I-135 present in the fuel, the second term on the right
hand side is initially zero. It remains negligible compared with the steady
production rate as long as the I-135 concentration is low. Thus, I-135 initially
builds up rapidly and then, as N
I
gradually increases, (so that λ
I
N
I
increases) the
net rate of growth of iodine falls (see Figure 5.2).
Eventually, N
I
increases to a high enough value that the decay rate is as high as the
steady fission production rate, and the net rate of growth becomes zero. I-135 is
then said to have reached its equilibrium concentration. In Figure 5.1, the outflow
from the upper box matches the inflow.
Notes:
CNSC 90
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

0
50
100
150
200
250
300
350
0 10 20 30 40 50 60
Time (hours)
I
o
d
i
n
e

R
e
s
e
r
v
e

B
a
n
k

(
m
k
)
xenon buildup at full power for comparison
iodine at full power
iodine at 60% of full power

Figure 5.2
Buildup of I-135 to Equilibrium Concentration
Under these conditions, the iodine equation becomes ) eq ( N 0
I I f I
λ − φ Σ γ = where
N
I
(eq) is the equilibrium concentration of I-135. Hence, re-arranging
φ
λ
Σ γ
=
I
f I
I
) eq ( N
This equation gives iodine concentration in the fuel in atoms/cm
3
. It is a little
easier to understand this expression if we convert it to iodine load in mk. For a
large CANDU this is: P mk 320 ) eq ( N
I
× − = where P, the neutron power, varies
between 0 and 1. The equilibrium concentration is proportional to flux (or power)
so, for example, the equilibrium iodine load is -320 mk at full power and at 50%
full power the equilibrium iodine load for the same reactor would be -160 mk.
Figure 5.2 shows the approach to this equilibrium value after reactor start-up. The
concentration reaches within 2% of its equilibrium value after 40 hours of reactor
operation. Note that the equilibrium level of I-135 is directly proportional to the
thermal neutron flux φ.
The equation for the buildup of I-135 as a function of time is well known
( ) ( )
t
I I
I
e 1 ) eq ( N t N
λ −
− ⋅ =
Because the exponential term in this equation is ½, ¼ etc. for successive half-lives
of iodine, the buildup reaches half way to equilibrium in one iodine half-life,
three-quarters of the way in two half-lives, seven-eighths in three, etc.

Notes:
CNSC 91
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Now we turn our attention to xenon buildup, which is more complicated
than iodine buildup. The initial xenon production rate, directly from fission, is
low. As soon as there is appreciable iodine, iodine decay increases the production
rate, and then xenon buildup tracks iodine buildup to equilibrium. The short initial
delay in xenon buildup is known as “hold-up”.
At equilibrium, xenon production equals loss (outflow matches inflow in the lower
box in Figure 5.1), so the condition for equilibrium is
| | | | ( ) eq N N 0
Xe Xe
Xe
a I I f Xe
λ + φ σ − λ + φ Σ γ =
where N
Xe
(eq) is the equilibrium concentration of Xe-135.
Xenon, produced mainly by I-135 decay, cannot reach equilibrium concentration
while the I-135 net production rate is changing. Since iodine reaches equilibrium
first, we can substitute the equilibrium concentration of I-135 into the xenon
equation for equilibrium. With this substitution, we find
| | | | ( ) eq N 0
Xe Xe
Xe
a f I Xe
λ + φ σ − φ Σ γ + γ =
Hence, re-arranging,
( )
φ
λ + φ σ
Σ γ + γ
=
Xe
Xe
a
f I Xe
Xe
) eq ( N
Again, we can convert this equation to xenon load in mk:
06 . 0 P 94 . 0
P mk 28
) eq ( N
Xe
+
× −
=
Figure 5.3 shows the buildup of Xe-135 to this equilibrium value of -28 mk over
about 40 hours; the upper curve is for the buildup at full power (P = 1), the lower
curve is for buildup at 60% full power (P = 0.6). Like I-135, Xe-135 reaches
within 2% of its equilibrium value after 40 hours of reactor operation.
The equilibrium Xe-135 concentration shown in Figure 5.3 is nearly the same at
60% and 100% full power. This is quite different than we noted earlier for the
equilibrium I-135 concentration, which is directly proportional to the power level.
Figure 5.4 shows the dependence of equilibrium xenon on reactor power.
Notes:
CNSC 92
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

0
5
10
15
20
25
30
0 10 20 30 40 50 60
Time (hours)
x
e
n
o
n

w
o
r
t
h

(
m
k
)
Full Power
60% of Full Power

Figure 5.3
Buildup of Xe-135 to Equilibrium Concentration
0.0
5.0
10.0
15.0
20.0
25.0
0.001 0.010 0.100 1.000
Fraction of Full Power
E
q
u
i
l
i
b
r
i
u
m

X
e
n
o
n

(
m
k
)
0.6
27.0
0.06 0.006

Figure 5.4
Equilibrium Xenon Load as a Function of Reactor Power
We can see why equilibrium xenon is nearly independent of flux for high flux by
examining the equation for N
xe
(eq). Taking the full-power flux for a large
CANDU reactor to be about φ = 10
14
n cm
-2
s
-1
, the sizes of the two terms on the
bottom line of the expression are:
1 5 14 24 6 Xe
a
s 10 35 10 10 10 5 . 3
− − −
× = × × × = φ σ and
1 5
Xe
s 10 12 . 2
− −
× = λ
At high flux, where the absorption rate is much bigger than the decay rate, we can
ignore the second term compared to the first. When we drop the decay constant
out of the denominator of N
Xe
(eq), flux cancels out and the equation becomes
independent of flux.
Notes:
CNSC 93
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This condition applies within the range of about 60% to 100% of full
power (see Figure 5.4).
Xenon Simulation
The reactor design gives it enough excess positive reactivity to overcome the
negative reactivity arising from equilibrium xenon. This means there could be a
large excess positive reactivity in the system when there is no xenon present,
which will be the case before the reactor has started to operate or on a restart after
a long shutdown—when the xenon has decayed. We must be able to compensate
for this excess reactivity. The usual way of offsetting the excess reactivity is to
dissolve a neutron absorbing poison (boron or gadolinium) in the moderator and
removing it as xenon builds up. The addition of poison to the moderator to
compensate for xenon that is “missing” is called xenon simulation.
5.2 TRANSIENT XENON BEHAVIOUR
Although the reactor has sufficient excess reactivity to offset equilibrium xenon, a
problem occurs following a shutdown after operation at power. Xenon reactivity
rises rapidly, peaking in about 10 hours. It takes nearly two days for xenon decay
to return the xenon load to near the full power equilibrium value. During this
time, there is not enough excess reactivity available to make the reactor critical so
it remains shut down.
The explanation of this effect, in broad outline, is not very difficult. Just before
shutdown, xenon removal (primarily by neutron absorption) matches the xenon
production (mainly from iodine decay). The large reservoir of iodine continues to
decay after shutdown, but xenon removal nearly stops. This imbalance produces
excess xenon.
To see in detail how this occurs, let us look back at the xenon equation. What we
have done below is to indicate underneath each term what its relative magnitude is
for a CANDU that has been running steadily at full power long enough to establish
equilibrium conditions (See Figure 5.1).
( ) | | | |
Xe Xe Xe
Xe
a f Xe I I Xe
N N N N
dt
d
λ + φ σ − φ Σ γ + λ =
90% 10% 90% 10%
At equilibrium, the production rate of xenon from iodine decay (λ
I
N
I
= γ
I
Σ
f
φ) is
much larger than its production rate directly as a fission product (γ
Xe
Σ
f
φ). At
equilibrium, the ratio between production from iodine decay and total
production—cancelling Σ
f
φ—is just γ
I
/(γ
Xe
+ γ
I
) = 0.064/0.070 = 0.91, hence the
90% to 10% (approximate) production ratio.

Notes:
CNSC 94
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Consider the xenon loss terms. The removal by burnout depends on the
value of the full power flux, so will be somewhat different for different CANDU
reactors. We have already established that under full power equilibrium
conditions for a large CANDU, the term λ
xe
N
Xe
is less than one-tenth of the term
σ
a
xe
φN
Xe
. (σ
a
xe
φN
Xe
≈ 10 × λ
xe
N
Xe
for φ = 0.6 of the full power flux.)
Consequently, burnup by neutron capture accounts, very approximately, for 90%
of the loss of xenon, while its own radioactive decay accounts for only about 10%
of the loss.
Consider now what happens following a reactor shut down after a long period of
operation at full power. The flux φ drops to a near-zero value within a minute or
so, removing direct xenon production almost immediately, but xenon production
by iodine decay continues. Net production remains near 90% of the equilibrium
value. On the loss side, we lose 90% of xenon removal, as burnup drops to zero,
leaving only radioactive decay. The result is that the xenon concentration starts to
rise quite rapidly, fed by the decay of iodine, as shown in Figure 5.5.
This cannot continue indefinitely, because there is a limited quantity of iodine in
the core and iodine production by fission stops following reactor shut down.
Iodine decays with its characteristic half live; half the iodine load decays in the
first half-life (6.7 hours), half of the remaining in the next 6.7 hours, etc.
Notes:
CNSC 95
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This generates a lot of xenon immediately after the trip, but less as time goes on
(see Figure 5.5). The xenon therefore reaches a peak value, (about 10 hours after
shutdown). Thereafter the xenon load gradually decreases, as the reduced rate of
iodine decay cannot keep up with the enhanced xenon decay.
0
20
40
60
80
100
120
140
160
180
200
0 10 20 30 40 50 60 70
time (hours)
x
e
n
o
n

l
o
a
d

(
m
k
)
Iodine
Xenon

Figure 5.5
Change in Iodine and Xenon Concentrations
Following a Trip from Full Power
Poison Prevent and Poison Override
Before we look at transients that are more complicated than a complete shutdown
from full power equilibrium conditions, we will explain the terminology shown in
Figure 5.6.
One function of the adjuster rods is to provide excess reactivity to override xenon
transients. Withdrawing adjuster rods from the reactor core contributes positive
reactivity, up to a maximum of 15 or 20 mk depending on the particular reactor. If
the negative reactivity due to xenon exceeds the adjuster reactivity worth, the
reactor is sub-critical with no way to restart it. We say it is poisoned out. Figure
5.6 shows that on a trip from full power, a reactor that is poisoned out cannot be
restarted until 35 or 40 hours after the trip, when xenon has decayed to near the –
28 mk equilibrium level.
Holding reactor power near 60% (or higher) allows sufficient xenon burnout to
prevent a poison out, (as Figure 5.8 shows). It is important to realize that on a
turbine trip it may make economic sense to keep the reactor operating and to
exhaust steam to a condenser (or the atmosphere). We call this mode of operation
poison prevent.

Notes:
CNSC 96
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Thirty minutes or so after a trip the negative reactivity from xenon exceeds
the reactivity worth of the adjusters (the poison override capability). (See
Figure 5.6 again.) A reactor restart and return to high power during the 30 minutes
or so of poison override time (the time from the trip to a poison out), rapidly burns
out the xenon and may prevent a poison out.
The adjusters drive out slowly, bank by bank, so, to avoid a poison out, the
operators must make the decision to try to restart and begin adjuster withdrawal
within about 20 minutes of the trip (the decision and action time).
Poison override is possible in principle, and is part of the reactor design, but is
often not practical. Before restarting the reactor following a trip, it is important to
find the cause of the trip and eliminate the fault. The operators must make a
number of checks before judging the trip “spurious” (that is, a trip that occurs
without an actual fault). Checks following a trip (or repairs) usually take longer
than the decision and action time, so operating procedures often do not allow the
operators to try to “beat the poison out”, removing the temptation to take a short
cut.

Figure 5.5
Xenon Reactivity Transient for a CANDU 600
The rate of rise of the xenon load after a trip is a function of the equilibrium
conditions before the trip. In CANDU reactors, the xenon load increase at about
0.5 mk per minute following a trip from full power. This number and the available
reactivity for poison override determine the poison override time. For example, if
a particular reactor has a maximum available reactivity of 16 mk, it must be
brought back to high power within (16 mk)/(0.5 mk/min) = 32 minutes.
Notes:
CNSC 97
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

25
30
35
40
45
50
0 5 10 15 20 25 30 35 40
time (hours)
x
e
n
o
n

w
o
r
t
h

(
m
k
)
straight line
xenon mk

Figure 5.6
The Initial Increase in Xenon Load Following a Trip
Example: Estimate the rate of xenon load increase following a trip for a reactor
that has an equilibrium iodine load of 300 mk.
The net rate of xenon production immediately following the trip is:
λ
I
N
I
(eq)- λ
Xe
N
Xe
(eq)
Using 300 mk and 28 mk for equilibrium iodine and xenon concentration gives
2.93 ×10
-5
s
-1
× 300 mk – 2.12 ×10
-5
s
-1
× 28 mk
= 8.79 × 10
-3
– 0.59 × 10
-3
= 8.2 × 10
-3
mk/s
= 0.5 mk/minute
Inevitably, this initial rate falls off, because the increase in xenon load makes the
product λ
Xe
N
Xe
(eq) bigger, while the depletion of the iodine load makes λ
I
N
I
(eq)
smaller. Figure 5.6 shows the first 40 minutes of increasing xenon load following
a trip to demonstrate that, during the override time, the rate of rise is nearly
constant.
Trips from Lower Power
Figure 5.7 shows trips from various power levels compared to a full power trip. In
every case, we are assuming equilibrium xenon and iodine loads before the trip.
The size of the xenon peak following a trip is nearly proportional to the iodine
load before the trip, and the equilibrium iodine concentration, as shown earlier, is
proportional to the pre-shutdown flux, so the height of the peak of the xenon load
is strongly dependent on flux level before the trip.

Notes:
CNSC 98
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Thus, for a trip from full power, the Pickering reactors have a xenon peak
of about 80 mk above the equilibrium xenon level, while the Bruce and Darlington
reactors have a peak 115 mk or so above equilibrium xenon. (All CANDU
reactors have nearly the same -28 mk worth of equilibrium xenon.)
0
20
40
60
80
100
120
140
0 10 20 30 40 50
time (hours)
x
e
n
o
n

l
o
a
d

(
m
k
)
xenon load (mk)
Trip from 100%
Trip from 80%
Trip from 60%
Trip from 40%
Trip from 20%

Figure 5.7
Xenon Transient Following Trips
from Various Equilibrium Power Levels
The time to the xenon peak is close to 10 hours for all high power trips, that is,
trips that occur with the initial equilibrium xenon close to 28 mk. Peaks following
trips from low power occur sooner.
After the peak passes, the xenon decay rate increasingly dominates the decrease in
xenon. (Twenty hours after the trip—about three iodine half-lives—the iodine
bank is down to about one-eighth of its pre-trip value.) As the iodine drops to
insignificant levels, the xenon decays with its half-life (t
½
= 9.1hours).
As you would expect, Figure 5.7 shows that a smaller initial iodine load makes the
poison override time longer and the poison out time shorter for trips from less than
full power.
Xenon Transients Following Power Changes
So far, we have discussed xenon transients occurring after a shutdown. The flux
goes quickly to zero, and only iodine decay and xenon decay remain as xenon
production and loss terms. In practical reactor operation, we are also interested in
the transients after a step reduction or a step increase in power. Solving the
corresponding xenon equations is a laborious chore and computer codes are
normally used.
Notes:
CNSC 99
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Nevertheless, you can explain the general shape of the curves by
comparing the sizes of the production and loss terms for xenon, just as we did for a
trip from full power. Any change in reactor power from the equilibrium steady-
state condition produces a transient change in xenon load because of the temporary
mismatch between the xenon loss and production terms. You can visualize the
changes in concentration as changes in “tank level” when the steady state “flows”
are changed in Figure 5.1
Figures 5.8 through 5.11 show the results of computer simulations of xenon for a
large CANDU reactor following various step changes in reactor power.
10
30
50
70
90
110
130
150
0 10 20 30 40 50 60
time (hrs)
x
e
n
o
n

l
o
a
d

(
m
k
)
FP to 20%
FP to 40%
FP to 50%
FP to 60%
trip from FP

Figure 5.8
Xenon Transients Following Step Power
Reductions from Equilibrium Full Power
Figure 5.8 shows the transients for 40, 50, 60, 80, and 100% power reductions
from initial full power. For a reduction of, say, 40% (that is, from full power to
60% F.P) the xenon removal by neutron capture also decreases by 40% from its
full-power value, but because xenon is still being removed by burnout, the
transient will not reach its shutdown peak. Looking at the figure, you will see that
for a reduction to 60% F.P., an available excess reactivity of ~15 mk would be
more than enough to override the transient altogether.
Ultimately, iodine reaches a new equilibrium at 60% of its full power value, and
the xenon reaches equilibrium with the xenon load a little less than -28 mk.
Figure 5.8 also shows that the rate of xenon buildup is less for a 60% reduction (to
40% F.P.) than for a trip. Reactivity peaks at a lower level a little sooner, and the
poison override time is longer.

Notes:
CNSC 100
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Figure 5.9 illustrates this in more detail. For a fixed amount of excess
reactivity, the poison override time depends on the size of the power reduction.
For example, the curve shows that the reactor will poison out in 50 minutes
following a power reduction to 30% full power, while a reduction to 40% full
power extends the override time to more than an hour.
Step Power Decreases from 100% Steady State
25
30
35
40
45
50
0 20 40 60 80 100 120
time (minutes)
x
e
n
o
n

l
o
a
d

(
m
k
)
trip FP to 0.1%
FP to 20%
FP to 30%
FP to 40%
FP to 45%
FP to 50%

Figure 5.9
Time to Poison Out vs. Size of Step Reduction from Full Power
Notes:
CNSC 101
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The converse of these curves also applies. Figure 5.10 compares the
change in xenon reactivity when power changes between 60% F.P. and 80% F.P.
For example, if the reactor is running at 60% F.P. (at equilibrium) and is taken to
80% F.P., the immediate effect is a gain in reactivity due to increased xenon
burnup. At the same time, the higher fission rate produces more iodine, but this
does not increase xenon production until later. Consequently, the xenon curve
runs through a minimum, and then xenon production rises because of the
increasing reservoir of decaying iodine.
20
25
30
35
0 5 10 15 20 25 30 35 40
time (hours)
w
o
r
t
h

(
m
k
)
80% to 60%
60% to 80%

Figure 5.10
Xenon Load Following a Power Increase to 80% from 60% FP
and a Decrease from 80% to 60%
The xenon stabilizes at 80% equilibrium load after the iodine has reached
equilibrium. The iodine reaches close to its 80% equilibrium value from its initial
60% following the usual half-life behaviour. It gets half way there (to 70%) in one
half-life, ¾ of the way (75%) in two half-lives. After 4 half-lives it is nearly at
79%. The iodine bank always proceeds smoothly in this way towards its new
equilibrium value. The xenon load reaches its new equilibrium value only after a
transient change in the opposite direction. This is caused by the immediate change
in removal by xenon burnout, coupled with the delayed change in production.
Notes:
CNSC 102
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

15
20
25
30
0.0 10.0 20.0 30.0
time (hours)
x
e
n
o
n

l
o
a
d

(
m
k
)
90% to 100%
80% to 100%
70% to 100%
60% to 100%
50% to 100%
Initial Buildup

Figure 5.11
Variation in Xenon Load Following Increase in Power to 100% FP
Figure 5.11 shows the change in xenon reactivity following an increase in power
to 100% from equilibrium operation at various lower powers. Normally, xenon
transients on power increases do not present particular operational problems, as
xenon simulation with moderator poison offsets the excess reactivity.
Reactor Restart with Xenon Present
If, following a trip, the reactor power increases within the poison override time,
(by adjuster removal) the burnout of xenon increases immediately. Consequently,
removal exceeds production and the xenon load peaks as power rises and then
comes down. Recall that the rate of iodine decay exceeds the rate of xenon decay
by about 0.5 mk/minute during the override time. Adding burnout to the xenon
decay brings the xenon back near equilibrium quickly. The burnout rate at full
power is, as we saw earlier, more than ten times the decay rate, so the “half-time”
for burnout is less than an hour.
Figure 5.12 shows the return to equilibrium following restart within the poison
override time. The curve assumes an instantaneous return to power, which makes
the peak unrealistically sharp. As the xenon burns down to the normal -28 mk
load, the regulating system will call for adjuster re-insertion, one bank at a time,
whenever the zones reach a high level.
Notes:
CNSC 103
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003



Figure 5.12
Restart During Poison Override
Return to power following a poison out (Figure 5.13) is a more usual operating
circumstance. Restarting the reactor after the xenon load has returned to about -28
mk but with almost all the iodine gone, produces a xenon removal rate (by decay
and burnout) that far exceeds the almost zero production rate.
Xenon simulation offsets the excess reactivity resulting from the rapid decrease in
xenon load. The transient positive reactivity looks like the Figure 5.11 curves
following power increase, with xenon returning to equilibrium following the return
of iodine to equilibrium.


Figure 5.13
Restart After a Poison Out
5.3 XENON OSCILLATIONS
Change in local flux causes a rapid change in xenon load, while the matching
change in xenon production is delayed by iodine buildup and decay. This can
cause a reactor to undergo periodic (i.e., repetitive) oscillations in flux level. A
suitably designed control system limits the size of the resulting flux peaks and
stabilizes the flux.

Notes:
CNSC 104
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

For the moment, ignore temperature reactivity feedback in the reactor and
look at xenon in isolation. Consider a small reactor operating at full power and
suppose that a small fluctuation takes place in the power with the effect of slightly
increasing the flux throughout the system. Because the flux increases, xenon
burnup increases and this increases the reactivity of the core, tending to produce a
further increase in the flux. There is therefore a positive feedback effect, which
pushes the power higher and higher unless some compensating action reduces
reactivity to counteract the reactivity increase.
In practice, the control system intervenes to prevent this continual rise. Even if the
control system is not entirely effective, the upwards drift would eventually stop
because iodine load gradually increases, producing more xenon and, eventually, a
decrease in reactivity. This would reduce the flux and therefore the burnup of the
xenon, magnifying the negative reactivity. The flux would then start to drop and
only recover when the reduced production of iodine at lower power resulted in a
lower input of xenon. In this way, the reactor power could continue to oscillate in
amplitude with a period of several hours between peaks. A control system that
reacts promptly to changes in power and maintains it at a more or less constant
level easily prevents this kind of bulk power oscillations. Temperature negative
reactivity feedback helps damp out small fluctuations.
For a small reactor, careful monitoring of bulk reactor power permits remedial
reactor power regulation and prevents oscillations. For a large reactor, however,
such as a CANDU, monitoring the overall power level is inadequate, because local
xenon oscillations can drive power up in one region of the reactor and down in
another, even though the total power remains constant.
Consider how localized oscillations occur. Suppose that the reactor is operating at
high power with a uniform power distribution, but the liquid zones in the top of
the reactor are a little above their nominal level, while at the bottom they are a
little below their nominal level. Although the control system is functioning as
intended to keep the flux distribution flat, this spatial configuration deviation from
the desired nominal configuration is sometimes referred to as a tilt. It may occur,
for example, after refuelling a few channels at the top without refuelling a similar
number of bottom channels.
Now, suppose that the regulating system continues to keep the total power output
of the reactor constant, but spatial control is lost and the zone levels return to
nominal. The flux increases a little in top of the core where the zone levels drop,
and simultaneously decreases in the bottom. This deviation of flux shape from the
normal flux distribution called a flux tilt. (In this particular example, it is a top-to-
bottom flux tilt). In the region of increased flux, xenon now burns out more
rapidly than it did before the change and its concentration decreases. The decrease
in xenon concentration leads to a higher flux, which again results in increased
local xenon burnup, increased local reactivity, increased flux and so on.

Notes:
CNSC 105
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Meanwhile, in the region of decreased flux, the xenon concentration
increases due to its reduced burnup and to the continued decay of the existing
iodine produced in the original, higher flux. This increased xenon concentration
decreases reactivity in this region, which reduces the flux, and in turn increases the
xenon concentration, and so on. The thermal flux, and hence the power density,
decreases in this region and increases in the other, while the total power of the
reactor remains constant.
These local power excursions do not continue in the same direction forever. At
the same time as the increased flux is causing xenon to burn out more rapidly in
the high-flux region, it is also increasing the production of iodine. The decay of
this enhanced iodine bank eventually leads to an increase in xenon concentration,
reducing reactivity and thus the flux and power in that region. Likewise, in the
region of reduced flux, the lowered production of iodine combined with the decay
of accumulated xenon increases the local reactivity and reverses the flux and
power transient in that region.
In this way, unless the regulation system responds to control them, the flux and
power may oscillate between different regions (top to bottom, end-to-end, or side
to side) indefinitely. Calculations show that xenon spatial oscillations have a
peak-to-peak cycle time of about 15 to 30 hours, and the height of the peaks may
increase from cycle to cycle.
Operationally, it is important to recognize that as the peaks turn into valleys and
valleys into peaks, there is a point in the cycle where the flux shape is normal.
Even though the xenon reactivity is (temporarily) the same in the two regions, the
oscillation is not under control. In the region where power is dropping, the iodine
load is near its peak and continues to force power down. Conversely, in the region
with rising power iodine load is at its lowest, so power continues increasing.
Conditions for Spatial Oscillations
This type of localized xenon oscillation can take place only in a large reactor, a
reactor whose spatial dimensions are large compared to the diffusion length of the
neutrons. With a small core, a disturbance started in one region affects other
regions because neutrons from the affected region spread the disturbance across
the core. As mentioned earlier, a regulating system that controls bulk power
adequately prevents oscillations in a small reactor.
When the dimensions of the reactor greatly exceed the distance travelled by the
thermal neutrons during their lifetime (which is the case in a large CANDU), a
disturbance that begins in one place does not spread its influence to a distant part
of the core, so the various regions act much more independently. Thus, if a flux
increase occurs in one region due to a fuel change, for example, a control system
based on maintaining the overall power constant will reduce the flux a little
throughout the core to compensate. This would set up a xenon oscillation in the
second region exactly out of phase with the one in the first region.
Notes:
CNSC 106
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The other condition that must pertain before spatial xenon oscillations can
occur is that the reactor operates at high power. When the flux increases at some
point in the reactor, the immediate increase in xenon burnup initiates the
oscillation. For a marked change in xenon concentration, xenon burnup must
predominate over xenon decay. We have already seen that this is the case for a
large CANDU, where xenon burnup at full power is at least a factor of 10 larger
than xenon decay. Typically, a spatial disturbance cannot become an oscillation in
a CANDU unless power is above 25% F.P.
The CANDU, then, like several other types of power reactor, satisfies the two
conditions required for spatial xenon oscillations to occur. Oscillations can occur
with constant (overall) reactor power, so they can continue unnoticed unless
instruments monitor the flux and/or power at several points throughout the reactor,
and localized absorbers respond to adjust local reactivity to counteract any flux
tilts.
Limiting Spatial Oscillations
Douglas Point at 200 MW(e) was the first CANDU to have regional absorbers to
control xenon oscillations; it was equipped with four absorbers, which controlled
four regions or zones in the reactor.
Pickering A at 540 MW(e) is quite a bit larger and so it is divided into 14 zones,
each with its own liquid zone control compartment and a pair of in-core flux
detectors. The regulating system uses the detector output to adjust the flow of
light water into the zone compartments. All later CANDU designs (Bruce,
Darlington, and Pickering B units and the CANDU 600) continue the use of 14
pairs of in-core detectors coupled with 14 liquid zone compartments.
Even with a zone control system, a severe xenon oscillation could risk significant
damage to the fuel. The size of the flux tilt depends on the size of the reactivity
upset that initiates the oscillation, and on the amount of compensating reactivity
that the zone control system provides. A large oscillation could drive one or more
liquid zones to their operating limits and spatial control would be lost in those
zones. Continued operation with a flux oscillation of such magnitude could lead,
at least, to a reactor trip or, more seriously, to dangerously high local fuel
temperatures or even fuel meltdown. Even without such severe consequences,
xenon oscillations burden the core materials with unnecessary temperature cycling
that could lead to premature materials failure.
5.4 SAMARIUM-149
Samarium-149 (often simply referred to just as samarium) has a large absorption
cross-section (42,000 barns) and a high production rate (total fission yield is
1.2%). Neither the cross section nor the yield is as big for this nuclide as for Xe-
135, so the samarium reactivity effects are much smaller.

Notes:
CNSC 107
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Fission does not produce Sm-149 directly. Sm-149 is the fission product
daughter of neodymium-149 and promethium-149. The decay sequence is as
follows:
γ + β + →

Pm Nd
149
61
149
60
t
½
= 1.7 h
γ + β + →

Sm Pm
149
62
149
61
t
½
= 53 h
Because of the short half-life of Nd-149 (compared to the half-life of Pm-149), we
consider the entire 1.2 % fission product yield to be promethium.
Like I-135, Pm-149 does not absorb neutrons; only decay can remove it. An
important difference between Xe-135 and Sm-149 is that samarium-149 is a stable
isotope, and therefore remains in the core after shutdown. Because it is stable,
only the process of neutron capture, with the reactor at power, can remove it.
Buildup of Pm-149 and Sm-149
We can write the relationships for rate of change of Pm-149 and Sm-149
concentrations in the same way we did for I-135 and Xe-135. You can easily
modify Figure 5.1 to apply to promethium and samarium.
Just like I-135, fission produces Pm-149 and decay removes it. The Pm-149
absorption cross-section is low enough that its rate of burnup (like 1-135) is
negligible compared to its rate of loss by decay. Consequently, the upper box in
Figure 5.1, with an appropriate change of labels (I → Pm) applies to Pm-149, and
the equation for the net rate of change of Pm-149 is identical in form to that for
I-135.
Promethium buildup, therefore, is similar to iodine buildup (½ way to equilibrium
in one half-life, ¾ in two half lives, etc.). Because of its long half-life, it takes
about 300 hours to reach equilibrium, compared to 40 hours for I-135. As with
iodine, equilibrium promethium concentration is proportional to flux.
Notes:
CNSC 108
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

0
1
2
3
4
5
6
0 100 200 300 400 500
time (hours)
S
m

L
o
a
d

(
m
k
)
buildup to 100%
buildup to 65%
buildup to 30%
buildup to 20%
Pm buildup to 100%
Prometheum
Samarium

Figure 5.14
Buildup of Pm-149 and Sm-149 in Fresh Fuel Following Startup
The expression for Sm-149 buildup is simpler than for Xe-135, because there is no
direct production of samarium from fission and no loss by decay. In re-labelling
the lower box of Figure 5.1 to apply to samarium, you need to drop the two
lightweight lines, the ones for direct production and decay.
Figure 5.14 shows the buildup to equilibrium of samarium and promethium in a
high flux CANDU reactor. The upper curve is for promethium, which builds up to
an equilibrium load (i.e., reserve inventory) of about -5.5 mk.
The hold-up for samarium is very much greater than we observed in Figure 3 for
xenon, and hold-up is more significant at low power. At high power, samarium
follows promethium buildup, reaching close to its equilibrium load of about -5 mk
in about 300 hours of operation.
Unlike xenon, the time required to reach equilibrium is a function of the flux level,
but the equilibrium samarium concentration is independent of the flux (for all
power levels). Another important difference from xenon is that samarium does not
decay after shutdown, so these samarium buildup curves (Figure 5.14), with no
initial samarium, apply only to fresh fuel inserted in the reactor.
Samarium Load After Shutdown
There is a transient rise in samarium concentration after a shutdown because
promethium decay continues, but burnup by neutron capture ceases when the flux
disappears. The maximum samarium load after shutdown depends on the
promethium load before shutdown, which depends on the reactor flux. For
CANDU reactors, the maximum samarium load after shutdown is between about 9
mk and 12 mk. Figure 5.15 shows the buildup. Note that the vertical axis starts at
the equilibrium samarium load.
Notes:
CNSC 109
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

5
6
7
8
9
10
11
0 7 14
Time (days)
S
m

L
o
a
d

(
m
k
)
Trip from 100%
Trip from 80%
Trip from 50%
Trip from 25%

Figure 5.15
Increase in Samarium Load After Shutdown
Although reactor design must allow for the equilibrium samarium load, the
shutdown load does not cause operational problems. There are two reasons for
this:
By looking at the time scale of Figure 5.15, you will realize that the maximum
samarium load appears long after the xenon peak decays. There will be lots of
reactivity available to deal with the samarium buildup when it occurs. The
transient rise in samarium is negligible during the xenon poison override-time, so
this does not present a problem either.
In addition to the excess reactivity available following xenon decay, there is a
plutonium-239 transient buildup that increases reactivity by a similar amount and
at about the same rate as samarium decreases it. The rate at which samarium is
formed after shutdown is governed by the Pm-149 half-life of 53 hours which, by
coincidence, is almost the same as the half-life of Np-239 (56 hours). Np-239 is
the parent of the Pu-239 that adds reactivity as it builds up in the fuel. After
shutdown, the Pu-239 starts to increase above its pre-shutdown value because
Np-239 decay continues, but Pu-239 burnup stops.
It turns out that the increased reactivity gained from the Pu-239 buildup more than
offsets the reactivity loss due to the increased Sm-149. The net result may be a
small reactivity gain of a couple of mk.
Samarium Load on Return to Power
Although samarium does not decay during shutdown, it will burn back to
equilibrium following a return to power. On restart, the promethium load builds to
equilibrium over 300 hours just as described earlier. The burnup rate of samarium
is significantly faster than this. It is also faster than the reduction of excess Pu-239
to equilibrium.
Notes:
CNSC 110
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The net effect is that, after xenon returns to equilibrium there is an excess
reactivity of several mk that disappears in a few more days. This does not cause
operational problems, unless normal fuelling stops (in the belief that there is
enough excess reactivity). If this happens, the core will require rapid refuelling
when the excess reactivity vanishes with the excess plutonium.
Similarly, transient changes in samarium on power level changes are very small
compared to xenon, and change very slowly over a week or so. In each case, the
samarium level returns to the same equilibrium value and the liquid zones easily
correct for small deviations from this value.

Notes:
CNSC 111
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

EFFECTS OF FUEL IRRADIATION AND ON-POWER
FUELLING ON CORE REACTIVITY
6.0 INTRODUCTION
During reactor operation, neutron reactions continuously change the composition
of the fuel. U-235 is burnt up, fissile Pu-239 and Pu-241 are produced (and burnt),
while neutron absorbing Pu-240 and fission product poisons build up. In CANDU
reactors, on-power fuelling compensates for the gradual decrease in net reactivity
that occurs as the fuel is exposed to neutron flux.
This section discusses the principal changes that occur in the fuel and examines
their magnitude and the time scale on which they are taking place. First, we
outline some characteristics of on-power fuelling. Then we define a unit that
specifies the degree of fuel burnup—the neutron per kilobarn. We will then look
specifically at the rates of U-235 consumption and Pu-239 build up at full power.
These studies enable us to consider how the changing composition of a fuel bundle
affects its contribution to the overall reactivity as irradiation proceeds. This
determines the fuelling rate that will keep the core critical.
6.1 ON-POWER FUELLING
CANDU reactors use on-power refuelling to maintain adequate long-term
reactivity. This keeps the amount of fissile material nearly constant by replacing
irradiated fuel with fresh fuel more or less continuously. While there are costs
associated with on-power fuelling (capital cost and maintenance costs of the
fuelling machines), this system has several distinct advantages over batch
refuelling:
• No downtime for refuelling;
• Easy removal of failed fuel, without a shutdown;
• Better average fuel burnup;
• Better flux shaping;
• It avoids the very large poison shim that is required if batch
refuelling is used.
In an equilibrium fuelled CANDU reactor, there are small local variations in flux
(peaks and valleys) associated with the state of burnup of the fuel locally, but, in
contrast to systems with batch refuelling, the overall reactivity of the core hardly
changes from day to day and month to month. The locations of the peaks and
valleys gradually change as daily fuelling proceeds, but the bulk properties of the
core remain the same. Section 8 describes how systematic fuelling controls local
flux peaks.
Notes:
CNSC 112
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

6.2 FUEL BURNUP—GENERAL
The initial fuel load of a new reactor is entirely fresh fuel, that is, fuel with no
plutonium or fission products present. Poison in the moderator compensates for
the excess reactivity of this fuel in the first few months of operation. Exposure to
neutron flux gradually changes the composition of the fuel, a process known as
fuel burnup. After four to six months of operation at high power, core reactivity
drops to a level where a poison shim is no longer required and routine replacement
of fuel becomes necessary to maintain core reactivity. The reactor is then
described as equilibrium-fuelled. Fuel is replaced on a daily basis (between 8 and
18 bundles per day) to add reactivity at a rate equal to its rate of loss from burnup.
These notes do not describe reactor operation or characteristics of the fresh-fuelled
core during the first months of operation. However, we do look at how the
composition of individual fuel bundles change during their time in the core.
In a freshly fuelled bundle, the only fissile material is U-235, which constitutes
0.72% of the natural uranium. Exposure to neutron flux gradually depletes the U-
235, decreasing reactivity. Buildup of the fission products, especially those with
significant absorption cross-sections for thermal neutrons further reduces
reactivity. These losses are only partly compensated by the buildup of fissile
Pu-239 following neutron radiative capture in U-238 (producing U-239, which
subsequently decays to Np-239, followed by a second beta decay to Pu-239).
Similarly, the net effect of Pu-240 and Pu-241 buildup is a net loss in reactivity.
Eventually, the gradual change in fuel composition requires replacement of the
high burnup fuel with fresh fuel to maintain the core critical.
Before reviewing the effects produced by fuel burnup, we should look at how to
track the reactivity contribution of an individual bundle, and at how to specify the
degree of burnup of fuel. The following two subsections describe these underlying
concepts.
Burnup Units
Three different units are in common use for describing the state of the fuel. Each
is described below.
Equivalent Full Power Days (EFPD)
Perhaps the simplest way to specify the burnup of a given fuel bundle is by the
number of equivalent full-power days (EFPD) it has resided in the core. This is
the number of days of exposure to full power flux. (A bundle exposed to 50% of
full power flux for two days would have a burnup of one EFPD). Documents for
non-specialists, such as public relations publications and management memos,
often use this measurement.
Energy Extracted per Unit Mass (MWh/kgU)
Notes:
CNSC 113
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Each watt of power production requires about 3.1 × 10
10
fissions per
second (see Section 1.1). One megawatt-day of heat energy production requires
the neutrons to fission about one gram of fissile material. One way to specify
burnup, then, is in terms of the total cumulative heat energy extracted from the
fuel. The unit is megawatt-hours per kilogram uranium (MWh/kgU). Accountants
or people involved in fuel purchasing are most likely to use this measurement.
Note that the MWh here are thermal, not electrical, energy.
Total Neutron Exposure (n/kb)
Since reaction rate is given by the product R = φΣ (Section 1.2), the rate of burnup
is proportional to the neutron flux. The accumulated burnup over a specific period
of time (t) is therefore proportional to the product of flux and time (φt). This
product is known as the total neutron exposure of the fuel; it is the fuel burnup
measurement of choice for reactor physicists, the fuelling engineer, and other
specialists. The units of neutron exposure are:
2
2
cm neutrons s
s cm
neutrons
t = × → φ
The total neutron exposures in these units are very big numbers, so a common
modification of the unit neutron/cm
2
is the neutron per kilobarn, defined by
changing the unit of area to the kilobarn:
1 kb = 10
3
b = 10
3
× 10
–24
cm
2
= 10
–21
cm
2
The relation between n/kb and n/cm
2
is then:
2 21
2 21
cm / neutrons 10
cm 10
neutron 1
kb / n 1 = =


The typical average exit burnup for fuel bundles discharged from a CANDU
reactor is around 1.8 n/kb. Individual bundles in the core at any given moment
have exposures that range from 0 to 2 n/kb or so.
One way to look at this rather arcane unit of exposure is to say that 1 n/kb is
equivalent to 10
21
n-cm per cm
3
, that is, the flux exposure measures the
accumulated track length of all neutrons that have passed through a unit volume of
the fuel during its time in the core.
The approximate relationship among the burnup units is:
100 MWh/kgU ≈1 n/kb =10
21
n/cm
2
≈ 115 EFPD

The units for energy extracted is not exactly proportional to the unit for neutron
exposure, as the energy extracted by a given flux exposure is slightly different for
Notes:
CNSC 114
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

fresh fuel and highly irradiated fuel. The conversion between EFPD and
the other burnup units above assumes a CANDU reactor with a full-power average
flux of 10
14
n cm
-2
s
-1
.
Tracking Fuel Burnup
Once a reactor has operated for some time, there is a wide variation in the
composition of the fuel bundles, since each will have a different degree of burnup
of U-235 and buildup of plutonium isotopes and fission product poisons. The
fuelling engineer keeps track of changes in individual bundles with a computer
simulation that calculates k

= ηfpε for each fuel bundle.
The fuel bundle reactivity is the difference (1 - 1/k

). It is really the reactivity of
an infinite reactor in which each unit cell contains a bundle with the same
composition as the bundle being tracked.
1.3
1.2
1.1
1.0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
F
o
u
r
-
f
a
c
t
o
r

p
a
r
a
m
e
t
e
r
s
Irradiation (n/kb)
0.9
η
ε
f
p
k

k

Figure 6.1
Variation with Burnup of the
Four-Factor Parameters, k∞ and k
The leakage factors (Λ
t
and Λ
f
) depend on the size and shape of the finite reactor.
The relation between k

and the effective multiplication factor k, is:
k = k

Λ
f
Λ
t

The k value in Figure 6.1 is about 30 mk below k

, indicating a value Λ
f
Λ
t
= 0.97.

Notes:
CNSC 115
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The equilibrium fuelled critical core with k = 1 consists of a large number
of bundles with different k

values determined by their individual degrees of
burnup. The lower k

values of the more highly irradiated bundles are
compensated by the higher k

values of the less irradiated ones. Each bundle can
be labelled with its k

value and its fuel bundle reactivity—i.e. its lattice cell
reactivity.
6.3 TRANSIENT REACTIVITY CHANGES
The concentrations of Pu-239 and its precursor Np-239 change transiently
following shutdown and restart (see Section 5.5). Steady operation produces an
equilibrium inventory of the precursor Np-239, which continues to decays
following shutdown. The decay to Pu-239 with a 56-hour half-life increases the
reactivity attributable to Pu-239 content by 12 mk over about 300 hours. On
restart, the excess Pu-239 burns back to equilibrium during the time that Np-239 is
rebuilding to its equilibrium value.
This section ignores all such transient effects associated with power changes or
shutdowns. We assume that once the fuelling machine inserts a fresh fuel bundle
into the core, the flux continuously irradiates the bundle at a steady high level.
Transient effects would appear “on top” of the long, steady changes in fuel
composition described here.
This section describes the initial buildup of plutonium isotopes and fission
products in the fuel, except for the initial buildup of xenon, which is not included
here. The initial 40-hour buildup of xenon produces a sharp 28 mk drop in
reactivity during the first 0.01 n/kb of flux exposure.
6.4 LONG-TERM REACTIVITY EFFECTS
The following subsections review the rate of burnup of U-235, the rate of buildup
of Pu-239, production of Pu-240 and Pu-241, the buildup of neutron absorbing
fission products, and the overall rate of reactivity change in the reactor due to the
changing composition of the fuel.
Burnup Rate of U-235
The fuel loses one U-235 atom whenever a U-235 nucleus absorbs a neutron. The
rate of absorption decreases as the U-235 concentration drops and this produces a
characteristic exponential drop in U-235, as Figure 6.2 shows. The removal rate
(slope of the curve) is highest when there is the most U-235, and decreases as U-
235 decreases.
Notes:
CNSC 116
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

N (t)
5
N
50
0
t

Figure 6.2
Exponential Decrease in U-235 Concentration
We can see how this behaviour occurs by looking at the removal process. Neutron
absorption removes U-235 from the fuel at the rate
5
a
5
a a 5
N R σ φ = Σ φ = We
write this as:
( )
5
5 5
N
dt
dN
a
φ σ − =
N
5
is the number of U-235 nuclei per cm
3
,
5
a
σ is the microscopic absorption cross-section of U-235 (in cm
2
),
φ is the neutron flux in n/cm
2
s.
The negative sign on the right hand side accounts for the fact that N
5
is decreasing.
This equation is of the same form as the radioactive decay equation:
N
dt
dN
λ − =
where the product (
5
a
σ φ) in the burnup equation is equivalent to the decay
constant λ in the decay equation, provided φ is constant. The U-235 concentration
therefore decreases in exponential fashion, as Figure 6.2 shows. It typically drops
to about ¼ of its initial value of 0.72% by the time the fuel is discarded.
The equation for U-235 concentration is
( )
( ) t
50 5
5
a
e N t N
φ σ −
=
where
50
N is the value of N
5
at time t = 0.
Notes:
CNSC 117
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Buildup Rate of Pu-239
Each neutron capture in a U-238 nucleus soon produces a Pu-239 nucleus. The
percentage change of the U-238 is very small (it constitutes about 99% of the fuel
regardless of whether it has been irradiated or not). While the loss of a small
amount of U-238 has no noticeable effect on reactivity, the production of a small
amount of fissile Pu-239 has a large effect.
The buildup of Pu-239, starting with fresh fuel, is quite similar to the buildup of
iodine. Figure 6.3 shows that it approaches an equilibrium value, reaching half
way to equilibrium in a certain interval, and then to ¾ in the same time interval,
etc. We can see how this behaviour occurs by looking at the production and
removal processes.
N
9
(t)
N
9
(eq)
0
t

Figure 6.3
Buildup of Pu-239
The steady state rate at which Pu-239 forms (ignoring hold-up—the time it takes
the two precursors to decay) equals the capture rate in U-238; or
φ Σ =
8
a a
R
where
8
a 8
8
a
N σ = Σ is the U-238 macroscopic absorption cross section. The U-238
concentration decreases only a little during irradiation so the macroscopic
absorption cross-section is almost constant. (The small thermal neutron absorption
cross-section
8
a
σ of 2.7 b causes a low burnout rate. Resonance capture is limited
by self-screening.)
Neutron absorption destroys Pu-239, resulting in either Pu-239 fission or capture
producing Pu-240. We can ignore Pu-239 removal by alpha decay because its
half-life is so long (over 24,000 years). The rate of loss of Pu-239 then equals:
( )
9
9
a a
N R φ σ =
Notes:
CNSC 118
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The net change in Pu-239 concentration per cm
3
per sec is:
( ) ( )
9
9
a
8
a
9
N
dt
dN
φ σ − φ Σ =
This equation is of the same form as the equation for iodine buildup, encountered
in Section 5. Initially, when there is no Pu-239 present, there is a constant
production rate. Once there is some Pu-239 present, the net production rate falls
off, because of Pu-239 burnout.
The buildup equation for Pu-239 is:
( )
(
¸
(

¸

− =
φ σ − t
9 9
9
a
e 1 ) eq ( N ) t ( N
Given sufficient time, the Pu-239 would build up to an equilibrium concentration.
( ) ( ) ( ) eq N 0
9
9
a
8
a
φ σ − φ Σ =
We can calculate this equilibrium value as a fraction of the U-238 content if we
replace
8
a
Σ with N
8
8
a
σ to give
( )
9
a
8
a
8
9
N
eq N
σ
σ
=
Substituting
8
a
σ = 2.7 b and
9
a
σ = 1013 b gives N
9(eq)
/N
8
= 0.27%, which compares
well with data presented in Table 6.1. The agreement is better than you might
expect, as we have ignored Pu-239 production by resonance capture in U-238 and
have ignored the strong temperature dependence of thermal neutron absorption in
Pu-239.
In practice, due to the overall loss of reactivity from burnup, it is necessary to start
replacing the fuel in the core well before the Pu-239 has reached equilibrium.
Pu-240 and Pu-241 Buildup
As noted earlier, some of the neutron absorptions in Pu-239 (about 27%) produce
Pu-240, which is non-fissile and has a large absorption cross-section for thermal
neutrons (290 barns). As significant amounts of Pu-239 begin to build up in the
fuel, Pu-240 also starts to appear. In principle, Pu-240 also eventually reaches an
equilibrium value, but actual fuel irradiation times are much too short for this to
occur. As far as we are concerned, Pu-240 is a poison that increases at a nearly
steady rate as the fuel burns up.
Pu-240 has one useful characteristic; neutron capture in Pu-240 creates fissile
Pu-241, with similar properties to Pu-239. This provides only inadequate
compensation for overall reactivity loss because only a small quantity of Pu-241 is
produced during the fuel irradiation time.
Notes:
CNSC 119
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Fission Product Buildup
The rate at which any given fission product builds up depends on the size of its
absorption cross-section. Through an argument similar to the one for Pu-239, we
can demonstrate that each fission product approaches an equilibrium concentration
at a rate that depends on its absorption cross-section.
The high cross-section fission products approach equilibrium relatively quickly.
They are known as saturating fission products. The two most important of these
are Xe-135 and Sm-149. The reactor design must include enough excess reactivity
to allow operation with saturating fission products present. The weaker absorbers,
the non-saturating fission products, build up gradually over long periods. Their
neutron absorption significantly affects the requirement to replace high burnup
fuel while it still contains significant amounts of fissile material.
Changes in Reactivity with Burnup
Table 6.1 lists the concentrations of the fissile isotopes U-235, Pu-239, and Pu-241
as a function burnup. Figure 6.4 plots these concentrations as a function of
burnup. The figure indicates the approximate exit irradiation for Point Lepreau,
Pickering A, and Bruce A fuel. The higher burnup achieved by the Bruce A
reactor is because it does not have adjuster rods in the core.
Notes:
CNSC 120
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003



Burnup Concentrations (g/kgU)
n/kb MWh/kgU U-235 Pu-239 Pu-241
0 0 7.20 0 0
0.2 19 6.37 0.60 0.002
0.4 39 5.62 1.10 0.009
0.6 59 4.90 1.48 0.025
0.8 79 4.30 1.77 0.049
1.0 100 3.76 1.98 0.078
1.2 120 3.32 2.14 0.107
1.4 140 2.90 2.25 0.145
1.6 159 2.56 2.33 0.177
1.8 179 2.26 2.39 0.211
2.0 198 1.98 2.43 0.245
2.2 216 1.74 2.46 0.278
2.4 235 1.54 2.48 0.309
2.6 253 1.35 2.49 0.338
2.8 271 1.18 2.50 0.366
3.0 289 1.03 2.50 0.393
* Strictly speaking, the values shown in this table apply only
to the Pickering reactors, but they will be correct to within a
percent or so for all natural uranium, D
2
O moderated reactors.
Table 6.1
Burnup Data*
Notes:
CNSC 121
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

7
6
5
4
3
2
1
0 0.5 1.0 1.5 2.0 2.5
1
2
3 4
1
2
3
4
PLGS outer core (256 channels)
PLGS inner core (124 channels) Bruce A outer core
Pickering A (whole core)
Bruce A inner core
C
o
n
c
e
n
t
r
a
t
i
o
n

(
g
/
k
g

U
)
Irradiation (n/kb)
Pu-239
U-235
Pu-241

Figure 6.4
Fissile Isotope Concentrations as a Function of Burnup
Figure 6.5 shows the more important components of the change of fuel bundle
reactivity (that is, the reactivity of the unit cell) with irradiation. Each fuel bundle
goes through the composition changes associated with the concentration changes
shown in Figure 6.4. Initially, as indicated by the combined curve (U-235 +
Pu-239), the positive reactivity contribution of the Pu-239 overcomes the negative
contribution due to burnup of U-235, so that the net effect is an increase of
reactivity. Initially, only eight Pu-239 atoms are produced for every ten U-235
destroyed, but the higher fission cross-section of the Pu-239 (742 b compared to
580 b for U-235) more than compensates for this ratio being less than one.
Eventually, the Pu-239 buildup rate slows. The fall-off in net Pu-239 production
means that, even though its concentration continues to increase, it can no longer
compensate for the continuing U-235 burnup, and the curve turns over.
The creation of Pu-240 produces a steady reactivity decrease, shown in the figure.
The buildup of fissile Pu-241 partly compensates for this reactivity loss. The
combined effect is a relatively steady decrease in reactivity.


Notes:
CNSC 122
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The neutron-absorbing fission products cause a negative reactivity that
builds up continuously as irradiation proceeds. The initial steep part of the curve
labelled “fission products” in Figure 6.5 is due to these, and particularly to Sm-
149. Note that the effect of the very strongly absorbing fission product xenon-135
is not included in Figure 6.5.
+20
+10
-10
-20
0 0.5 1.0 1.5 2.0
Change in
Reactivity (mk)
Irradiation (n/kb)
Fission Products
U-235 + Pu-239
Pu-240
Total
Pu-241
Typical discharge
burnup

Figure 6.5
Components of Change of Reactivity with Burnup
The net effect of all these contributions is shown by the “total” curve in Figure 6.5.
A comparison with Figure 6.1 shows it is essentially the same as the curve for k

.
The factor that describes the observed changes in k

is the reproduction factor (η).
We recall from Section 1.4 that:
( )
( ) fuel
fuel
a
f
Σ
Σ
ν = η
The change in “fuel bundle reactivity” (by which we mean the reactivity of a unit
cell containing such a bundle), is determined by how the fission and absorption
cross sections in the fuel change due to changes in concentration of fissile isotopes
and fission products.
The sharp initial dip in reactivity comes from the accumulation of the saturating
fission product Sm-149, which builds to near its equilibrium value over the first
week or so of operation. Sm-149, with its high thermal neutron cross-section,
increases the absorption cross-section significantly without affecting the fission
cross-section. Once this reaches equilibrium, the initial Pu-239 buildup more than
compensates for U-235 burnup.
Notes:
CNSC 123
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The reactivity increases because the change in concentration of the fissile
materials affects the numerator of η more than the denominator. The net positive
contribution to fuel bundle reactivity continues until the Pu-239 growth rate falls
off as it moves toward equilibrium, and no longer compensates for continuing
U-235 burnup. The peak in the total reactivity curve (at an irradiation near 0.5
n/kb) is known as the plutonium peak.
The steady buildup of non-saturating fission products and Pu-240 in the fuel
makes the decrease in reactivity steeper. Buildup of Pu-241 reduces that rate of
reactivity loss somewhat, but is unable to reverse it. The steady buildup of
Pu-240/Pu-241 and the non-saturating fission products combine to increase the
denominator of η, with only a small increase in the numerator from Pu-241.
At some point, therefore, it becomes necessary to start removing partially burnt-up
fuel, replacing it with fresh fuel, well before the fissile content is used up.
Section 8 considers the practical aspects of choosing channels for refuelling.
The total reactivity curve crosses the axis (reactivity = 0) near one n/kb. The
equilibrium fuelled core contains a mixture of bundles with reactivity (k

) values
above and below this irradiation. In the equilibrium-fuelled core, reactivity drops
steadily with irradiation. The CANDU 600, for example, consumes about 0.4 mk
per day (that is, the reactivity worth of the fuel diminishes at that rate). The slope
of the curve near k = 1 in Figure 6.1 determines the rate of loss of reactivity, which
determines the refuelling rate. To restore 0.4 mk of reactivity each day by
refuelling requires replacement of some fuel in each of 2 to 3 fuel channels each
day.
The average burnup of fuel discharged from the reactor is larger than one might
expect from the graphs, which show k = 1 for an irradiation of 1.1 n/kb. The
figures assume a reactor core that is burning up uniformly, while the equilibrium-
fuelled reactor contains a mixture of fuel bundles of differing degrees of
irradiation. Some of these bundles (the ones nearing discharge) have an irradiation
appreciably greater than 1.1 n/kb, but fresh bundles, with irradiation below 1.1
n/kb, compensate for the reactivity “deficit” of the high burnup bundles. The
efficiency in using the fissile material in the fuel is generally reported by giving
the discharge burnup (sometimes called the exit burnup) which is likely to average
1.8 n/kb or more.
If the refuelling machines are unavailable for some reason, the reactor can
continue to operate for only a limited time. For example, if the liquid zone
compartments were all 50% full, about 2 mk of excess reactivity would be
available. This would allow nearly five days of operation by gradual reduction of
liquid zone levels.



Notes:
CNSC 125
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

REACTOR OPERATIONS AT LOW POWER
7.0 INTRODUCTION
This section combines reactor dynamic behaviour, discussed in earlier sections,
with reactor operations at low power. We begin with some power definitions and
then describe the power rundown of an equilibrium fuelled CANDU reactor that
has been operating at steady full power. We will describe the rundown of neutron
power in terms of its dependence on prompt neutrons, delayed neutrons, and
photoneutrons and compare the neutron power rundown with the much slower
decrease in thermal power caused by fission-product decay heat. We then look at
the shutdown state and discuss procedures for approaching criticality. Finally, we
look at a reactor operated in the low power (almost) critical state and review
feedback effects that occur as power rises into the last decade or so of reactor
operation.

7.1 THERMAL POWER, NEURON POWER, AND FISSION POWER
The power referred to most frequently in this course is neutron power, which is
essentially the fission rate multiplied by the prompt energy release per fission. We
cannot measure this directly, but we do monitor the flux and the average neutron
flux in the core is proportional to the overall fission rate.
Reactor thermal power (often shortened to thermal power) is the reactor’s rate of
heat energy production. Thermal power takes account of nuclear decay heating
and of conventional heat (pump heat and heat losses to the environment).
Fission power is the name given to the heat generated because of nuclear processes
in the fuel. This includes wasted heat, such as that generated in the moderator and
shielding. It does not include any conventional heating. The operating license
places an upper limit on fission power, which is enforced by regulating the reactor
thermal power.
Notes:
CNSC 126
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Figure 7.1 illustrates the relationship among these powers for a CANDU 600. The
numbers will differ from reactor to reactor, but the per cent values are quite
similar.
Reactor Power
Heat to Coolant
Fission
Power
Neutron
Power
DECAY
HEAT
HEAT TO
MODERATOR
AND
SHIELDING
PUMP
HEAT
PIPING
LOSSES
HEAT PRODUCED IN THE FUEL
LICENSE LIMIT — 2156 MW(th)
(typical value for CANDU 6)
5%
6% 89%
0.3%
1%
6.3% 93%
THERMAL POWER TO SECONDARY SIDE
This is what we want the regulating system to control
Nominal 2064 MW(th)

Figure 7.1
Approximate Heat Balance for a CANDU 600 Showing Fission Power,
Thermal Power, and Neutron Power
7.2 REACTOR POWER RUNDOWN
The fission reaction is responsible for the power generated in a nuclear reactor,
and the fission rate is proportional to the neutron flux, so you might expect that
thermal power output is proportional to neutron power. The relationship between
flux and thermal power, however, is not linear. For example, if we drop from
100% to 10% neutron power, as indicated by flux-measuring instruments, the
thermal power does not fall to 10% of its full power value. Even though the
neutron power has decreased by a factor of 10, contributions to thermal power of
decay heat, pump heat, and ambient losses do not change, or change slowly.
These three sources of non-linearity behave as follows:


Notes:
CNSC 127
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The heat produced by the radioactive decay of the fission products in the
core (fission-product decay heat) changes slowly following a power change. After
sufficiently long reactor operation, 6% to 7% of the steady state heat generated
comes from the inventory of fission products. The heat comes from the beta and
gamma decay of these products rather than directly from the fission process itself.
Thus, if the reactor has operated at 100% power for a long time and is then shut
down, even though the fission process stops more or less instantaneously, the
thermal output immediately after shutdown is still 7% of its full power value, and
will decrease slowly as the fission products decay.
Figure 7.2 shows how the reactor thermal power and the neutron power decrease
after a shutdown. Note that after a minute or so, the neutron power is contributing
only a small fraction of the total power generated in the core.
Power From
Decay Heat
Thermal Power
in the Fuel
Measured Neutron
Power
Note Supressed Zero
3 Minutes 1 Hour
0.1 1 10 100 1000
0.1%
1%
10%
100%
Power
(%FP)
Time After Reactor Trip (s)

Figure 7.2
Decay of neutron and thermal power after shutdown
The same thing happens, on a less dramatic scale, whenever there is a change from
steady operation at one power level to another. When the reactor shifts from 50%
to 100% of full power, the ratio of decay heat to neutron power drops to roughly
half its previous level, and then builds up slowly as the fission product inventory
adjusts to the new power.
Another source of non-linearity is heat generated by fluid friction. About two-
thirds of the pressure drop in the heat transport system occurs in the turbulent flow
of the coolant channels. This means that about two-thirds of the heat input of the
heat transport pumps appears as heat in the coolant channels (about 13 MW(t) at
Bruce-A). This input depends exclusively on the coolant flow rate, so it is
independent of the reactor power level. This heat, known as pump heat, continues
as long as the main pumps are operated.

Notes:
CNSC 128
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

A third source of non-linearity is the heat lost from the coolant channels,
mainly to the moderator (about 4 MW(t) at Bruce-A, for example). The amount of
heat lost depends on the temperature difference between the coolant and the
moderator, which changes very little over a wide range of reactor power.
We will now take a detailed look at the decay of neutron power after a shutdown.
The section ends with a quantitative description of the much slower thermal power
rundown.
Neutron Power Rundown
To make things concrete, let's assume that an equilibrium-fuelled reactor has been
running at full power and is then shut down by a trip which inserts -100 mk of
reactivity, giving k = 0.90. Figures 7.3 and 7.4 show the time variation of neutron
power following shutdown.
Notes:
CNSC 129
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

0.001
0.01
0.1
1
10
100
0 1 2 3
Time (minutes)
N
e
u
t
r
o
n

P
o
w
e
r
(
%

F
.
P
.
)
observed power
total source
photoneutron source
delayed neutron source
Region I
Region II

Figure 7.3
Neutron Power after Shutdown (First 3 minutes)
0.00001
0.0001
0.001
0.01
0.1
1
10
100
0 1 2 3 4 5 6 7 8 9 10
Time (minutes)
N
e
u
t
r
o
n

P
o
w
e
r
(
%

F
.
P
.
)
observed power
total source
photoneutron
source
delayed neutron source
Region I
Region
II
Region III

Figure 7.4
Neutron Power after Shutdown (10 minutes)
It is convenient to divide the neutron power rundown curve into three regions and
explain them one at a time in the following subsections.
Notes:
CNSC 130
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Region I – The Prompt Collapse
With k = 0.90, the original prompt neutron population will decrease initially by a
factor of 0.90 every generation. For example, in 0.5 seconds there is time for 500
prompt neutron lifetimes of 0.001 s each. The neutron power would drop to (0.90)

500
≈ 10
-23
of full power, if there were no delayed neutron precursors continuing to
produce neutron. This corresponds to the very rapid initial drop shown in Figure
7.3. The collapse of the prompt population in Region I is so fast that, in practical
terms, it collapses at the rate at which negative reactivity is inserted into the core.
Region II – Delayed Neutron Hold-up
As we know from Section 3, the neutron population does not continue dropping at
this rapid rate, but stabilizes temporarily at a level determined by the subcritical
multiplication of the delayed neutrons in the reactor. There are two ways to
estimate where the prompt drop will “stop”.
In the first, we simply use the prompt jump formula from Section 2. The power
immediately after the drop is:
0
P
k
P
∆ − β
β
=
which, for an injection of -100 mk into an equilibrium fuelled core (β = 0.005),
gives:
0 0
P 048 . 0 P
100 . 0 005 . 0
005 . 0
P =
+
=
that is, the insertion of -100 mk of reactivity causes neutron power to fall
immediately to about 5% of its original value.
Alternatively, we note that when the reactor is critical at a steady power level, the
thermal neutron population originates from prompt and delayed neutrons, with a
fraction 0.995 prompt and the remaining 0.005 delayed. The prompt fraction
disappears very rapidly on shutdown, before the delayed precursors have had time
to decay. The power would drop to 0.005 P
0
(0.5% of full power) without
subcritical multiplication. However, the delayed neutron precursors decaying in
the now subcritical reactor can be regarded as a source with initial source strength
P
source
= 0.005 P
0
.


Notes:
CNSC 131
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Because of the subcritical multiplication factor l/(l-k) = 1/0.100 =10, the
actual power level immediately after the prompt neutron population disappears is:
( )
0 0 source
P 05 . 0 P 005 . 0 10 P
k 1
1
P = × = ⋅

=
This agrees with the value calculated from the prompt jump formula. Either
method is an acceptable approximation.
The “delayed neutron source power” (the power that would be produced by the
delayed neutron source alone, in the absence of subcritical multiplication) falls off
as shown in Figure 7.4. Initially the drop is rapid as the shorter-lived precursors
decay, and then it slows until eventually the decrease is governed by the longest-
lived precursor group. The delayed neutron source power is negligible (compared
to the photoneutron source) less than ten minutes after shutdown.
Region III – Photoneutron Hold-up
In the critical reactor, the photoneutron source is considerably weaker than the
delayed neutron source. However, the half-lives of the fission products whose
gamma rays produce the photoneutrons are generally much longer than the half-
lives of the fission products that generate the delayed neutrons. After a trip, the
photoneutrons become the dominant neutron source in the reactor within two to
three minutes (see Figure 7.4).
Table 7.1 lists nine photoneutron groups with the initial strength and half-life of
each group. For equilibrium fuel, the photoneutron fraction is about 0.033%.

Group
Fraction of
0.033%
t
½

9 0.646 2.5 s
8 0.203 41.0 s
7 0.070 2.4 m
6 0.033 7.7 m
5 0.021 27.0 m
4 0.023 1.65 h
3 0.003 4.4 h
2 0.001 53.0 h
1 0.0005 12.8 d
Table 7.1
Photoneutron Data
Figure 7.5 shows the gradual falloff in photoneutron strength after a shutdown. In
a week or so, the only significant source strength comes from the group 1
photoneutrons. This photoneutron source has a half-life of 12.8 days and keeps
the reactor power measurements on scale for several weeks after shutdown.
Notes:
CNSC 132
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Eventually (after a year or more) the photoneutron strength decreases until
it is comparable to the spontaneous fission source strength (≈ 10
-14
F.P.). The
photoneutron source strength continues to drop, but the spontaneous fission source
is constant, so the power will stop decreasing at a value of about
(1 - k)
-1
× 10
-14
F.P.
1.E-07
1.E-06
1.E-05
1.E-04
1.E-03
1.E-02
1.E-01
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14
WEEKS
S
O
U
R
C
E

S
T
R
E
N
G
T
H

(
%

F
.
P
.
)

Figure 7.5
Photoneutron Source Strength Following Shutdown
Two points are worth emphasizing about the decay of neutron power:
First, although we have divided the rundown into three distinct regions, in practice
the transition between Regions II and III is arbitrary. In Region II, both the
delayed neutrons and the photoneutrons contribute to the source, but the balance
between the two shifts steadily towards the photoneutrons as time goes on.
Second, throughout the entire rundown, the great majority of neutrons that appear
in the reactor are prompt fission neutrons from the fuel. Although, in the
subcritical reactor, fission would not take place in the absence of a source of some
kind, it is not the source, but the assembly itself that produces most of the neutrons
because of its subcritical multiplication. In the case illustrated, where k=0.90,
each gamma ray that generates a photoneutron gives rise to an average of 10
neutrons [1/(1-k) = 1/0.1 ] because of subcritical multiplication.
Thermal Power Rundown
Following shutdown, the thermal power of the reactor decreases much more
slowly than the neutron power, mainly because of the decay heat associated with
fission products in the fuel. Decay heat produces nearly 7% of the steady state
thermal power. Although the fission rate falls off rapidly, decay heat can only fall
off at the decay rate of the fission products producing it.
Notes:
CNSC 133
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Fission products have half-lives ranging from fractions of a second to
thousands of years. The longer-term fall-off in thermal power is therefore a very
slow one.
Typically, thermal power drops to about 3% full power in about three minutes
(which explains, incidentally, why 3% is the capacity chosen for the auxiliary
boiler feed-pump), to about half of this (1.5%) in an hour, and to below 1% over a
period of eight hours.
As a practical point, we should note another factor that slows the rate of decay of
thermal power generated in the core. The heat transport pumps generate heat at
the rate of nearly 1% of full power. This heat source persists until it is possible to
switch to the much smaller shutdown cooling pumps.
7.3 THE SHUTDOWN STATE
During operation, a reactor is always critical or in the deeply subcritical
guaranteed shutdown state (GSS), or in active transition between these states. The
observed steady-state power in a subcritical reactor is always given by the
subcritical multiplication formula:
source obs
P
k 1
1
P ⋅
(
¸
(

¸


=
After the reactor is placed in the GSS following extended high power operation,
the observed power level changes because both k and P
source
are changing.
Although the poison concentration may not be changed while in the GSS,
temperature changes and the decay of fission products both affect core reactivity.
Decay of photoneutron precursors gradually changes the neutron source.
Reactivity Changes
A number of inherent reactivity changes in the reactor core follow shutdown from
extended high power operation. The effect of the power coefficient is the most
immediate. Then, following the xenon transient and decay of xenon, other smaller
effects appear. Table 7.2 lists the effects and gives typical CANDU values for
each and the time over which the reactivity changes. The range of values in the
table is intended to encompass the real variations from reactor to reactor and
uncertainties in some of the parameters.
The net effect of these changes is that when the reactor returns to the hot shutdown
state ready for restart, the core reactivity is +30 to +32 mk higher than it was
before shutdown. Moderator poison keeps the reactor deeply subcritical. On
restart, excess poison is removed and the poison concentration at criticality is
exactly that needed to offset the increase in reactivity that has occurred during
shutdown.
Notes:
CNSC 134
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

For example, if the core reactivity increases by 31 mk, the poison
concentration at critical is equivalent to about 1.1 ppm Gd
(1.1 ppm × 28 mk/ppm = 31 mk). The reactivity increase must account for any
shutdown fuelling, any change in reactivity device configuration, or any
temperature differences as well as the effects in Table 7.2.
Off-line calculations to predict these changes in reactivity are unlikely to be more
accurate than ± 0.3 mk (1% of 30 mk). In practice, predictions of critical poison
concentration are often no better than ± 1 mk, equivalent to ± 15% zone level or
±0.04 ppm Gd. The reason for mentioning this is to point out that it is not possible
to know the exact poison concentration at criticality in advance. (The problem of
accurate poison concentration measurements compounds the uncertainty.)


Reactivity Effect
(mk)
Time Scale
Temperature Effects
Full Power Hot
to Hot Shutdown
+3 seconds to minutes
Hot Shutdown
to Cold Shutdown
-4 to -6 minutes to hours
Fission Product or
Precursor Decay

Xenon Decay +28 80
+
hours
Np-239 Decay +9 to +12 2 weeks
Pm-149 Decay -8 to -11 2 weeks
Other Saturating
FP Decay
-0.5 to -1 5 to 10 days
Table 7.2
Reactivity Changes Following Shutdown
Power Level Changes in a Shutdown Reactor
Once the reactor is cooled, placed in the GSS, and xenon has decayed (after 3 days
or so), there are only small reactivity changes in the shutdown core. The observed
power then trends downward with the gradual decrease in the source strength.
Given a good estimate of core reactivity, we can determine the source strength
from the measured power. Suppose, for example, that the reactor is subcritical by
400 mk (k = 0.6). The subcritical multiplication factor is 2.5 (1/0.4 = 2.5) so the
observed power level is two and a half times the source strength. Figure 7.5 shows
the neutron source trending downward during the three months following a
shutdown. (Source strength estimates beyond about three months have been
unreliable.)
Notes:
CNSC 135
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

7.4 APPROACH TO CRITICAL
We will consider two somewhat different approaches to critical from the
guaranteed shutdown state (GSS):
• Startup with the reactor regulating system instruments initially off
scale low.
• Startup with reactor regulating system instruments on scale.
We do not discuss initial startup or restart after a very long shutdown (~ 1 year or
longer) in which spontaneous fission neutrons are the source neutrons. These are
“once in a lifetime” events, for which training is given as required.
Reactor power monitoring instruments go off scale low some time after shutdown.
The time, typically a few weeks, depends on the particular reactor, its operating
history, and the depth of its GSS.
Example:
Suppose your reactor is 250 mk subcritical in the guaranteed shutdown state and
your RRS instruments go off-scale low at an observed power level of 4 × 10
-6
F.P.
The subcritical multiplication factor is 1/0.25 = 4, so the instruments go off scale
low when the source power is 10
-6
F.P. From Figure 7.5, this is about seven weeks
after shutdown.
Repeating this calculation with a reactor that is 667 mk subcritical in the GSS, and
with RRS instruments that go off-scale at 9 × 10
-6
F.P. we find the instruments go
off-scale low about three weeks after shutdown.
Any restart from the GSS requires manual operation of the purification system to
remove moderator poison. The monitoring instruments differ depending on the
length of shutdown. For restarts within a few weeks of shutdown RRS
instruments monitor the startup; for restarts after an extended outage, special
startup instruments monitor the initial stages of startup, with a transfer to the RRS
instruments after they come on scale.
Before looking at these two types of start-up, we will review some principles
common to both. First we must discuss what we mean by critical. Then we will
look at how the core sensitivity changes in going from the deeply subcritical GSS
to the almost critical state.
Criticality
With a neutron source present, there is always a steady state subcritical power
level associated with each value of the effective multiplication constant for k < 1.
For example, if the reactor is 1 mk subcritical, the subcritical multiplication factor
is 1000.
Notes:
CNSC 136
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

You could draw a parallel curve in Figure 7.5 three decades higher than
the source strength shown and know that whenever the observed power falls below
that curve the reactor is more than 1 mk subcritical (assuming the predicted source
strength is correct for the particular reactor).
We usually declare the reactor critical when, technically, it is not quite critical.
The criterion is whether the regulating system has the capability of taking the
reactor slightly supercritical (to manoeuvre power at the demanded rate) without
excessively draining the zones. When this condition pertains, the reactor is said to
be under direct regulating system control, or operationally critical. Station
procedures may differ, but a typical standard is to declare the reactor “critical” (or,
under direct regulating system control) when a 10% decrease in zone level (or
less) can achieve a subcritical power doubling.
Example:
Recall from Section 3, that power doubles when the subcritical reactivity is
halved. If a 10% zone level drop corresponds to adding ~ 0.6 mk, then, by this
measure, the reactor is considered critical when ∆k = -1.2 mk (k = 0.9988). After
the observed power doubling, it is subcritical by 0.6 mk.
Power Increase in the Subcritical Core during Startup
We can estimate the power change that will occur when the reactivity changes
from deeply subcritical to almost critical without knowing the source strength. For
example, applying the subcritical multiplication formula to the initial and final
states, the ratio of observed power for k nearly equal to one and in the GSS is:
( )
( )
" critical "
GSS
obs
obs
k 1
k 1
GSS P
" critical " P


=
Suppose that the reactor is subcritical by about 600 mk in the GSS. (This takes
into account the mk worth of poison added following shutdown, and the
underlying changes in inherent core reactivity discussed in Section 7.3.) Assume
that the reactor is considered critical when ∆k = - 0.6 mk. Then:
( )
( )
1000
0006 . 0
600 . 0
GSS P
" critical " P
obs
obs
= =
For this reactor, power will rise by three orders of magnitude during the startup.
Notice that the power level at which the reactor goes “critical” is likely to differ
from one startup to the next. It depends on the power level observed prior to
startup, which depends on the source strength. Using Figure 7.5 we can see that
after a short shutdown (less than one week) the reactor reaches criticality above
10
-4
F.P. After a long shutdown of, say, twelve or thirteen weeks the power at
critical will be a little above 10
-6
F.P.
Notes:
CNSC 137
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The responsiveness of a subcritical core depends on how close the reactor
is to critical, not on the power level. A reactivity addition has a much larger effect
if the reactor is nearly critical than it does in a deeply subcritical core (see
Section 3). The ratio of observed power (count rate for start-up instruments)
before and after the addition is:
( )
( )
f
i
obs
obs
k 1
k 1
initial P
final P


=
Suppose the reactor is 500 mk subcritical (1-k
i
) = 0.5 and we add +100 mk of
reactivity, so (1-k
f
) = 0.4. The power would increase by 25% (0.5/0.4 = 1.25). In
contrast, adding only +1 mk to a reactor that is initially 2 mk subcritical doubles
the power. [In this case, (1 - k
i
) = 0.002 mk and (1 - k
f
) = 0.001 mk.]
Clearly, the sensitivity of the core to reactivity insertions changes dramatically as
the reactor gets closer to critical. The dynamic response also changes
dramatically.
A step increase in reactivity when the reactor is deeply subcritical produces an
immediate change in subcritical power. The observed power change keeps up
with the reactivity changes during continuous poison removal. When the poison
removal stops, the power increase stops.
In contrast, with the reactor almost critical, it may take several minutes for power
to reach the new equilibrium power level following each step removal of
reactivity. During continuous poison removal, the power rise lags behind the
reactivity change. Even if the poison removal stops, power continues to rise for
some time. Table 7.3 indicates the stabilization time for different core reactivities.

Core Reactivity
(mk)
Stabilization Time
3
(3 × ,τ,)
- 100 35 s
- 10 50 s
- 4 1 ¼ m
- 2 2 m
- 1 3 ½ m
- 0.5 6 m
- 0.25 12 m
- 0.1 30 m
Table 7.3
Stabilization Following a Step Reactivity Insertion (Subcritical Core)


3
Based on the decay constant for the second longest lived delayed neutron group
Notes:
CNSC 138
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Initial Approach to Criticality after an Extended Outage
During long shutdowns, where the power drops to a low level, the normal
instrument readings are unreliable because background gamma rays contribute to
the readings. The installed instruments are considered “off-scale” somewhere
between about 10
-6
and 10
-7
of full power. The readings below this level are not
proportional to the flux. Supplementary “startup instruments” (BF
3
counters—He-
3 counters at some stations) are installed before this happens, and the initial stages
in the subsequent approach to critical would use these counters.
The initial approach to criticality following removal of the GSS uses the
purification system to increase reactivity and the startup instruments to monitor
count rate. As there is no automatic regulation, the operating staff must perform
the functions of the automatic system. This means observing power level, making
sure that changes in power are within the expected range, and adjusting
purification flow appropriately. Monitoring usually includes plotting a graph of
1/CR vs. reactivity (or vs. poison concentration), as described near the end of
Section 3.
Some stations use a continuous poison pull and plot 1/CR vs. time. The ion-
exchange columns have a half time for removal of gadolinium. That is, they will
remove half the poison in the moderator in a certain number of hours, half again in
the same amount of time, etc. In other words, poison removal slows as poison
concentration decreases, so a graph of 1/CR vs. time is not linear.
For either the linear or the non-linear graph, it is important to know the expected
trend and to monitor routinely to verify that the actual trend matches the expected
one. At some point in this process, the normal instruments come on scale. The
operator then uses the RRS ion chambers, the liquid zone control system, and the
purification system to take the reactor critical.
Approach to Critical Using RRS
With the normal RRS ion chambers in range, the regulating system responds, but
the reactor is not close enough to critical to say the regulating system is in control.
The effect of valving in purification with the regulating system holding power is to
raise the zone levels with the net reactivity of the core unchanged. With
purification valved out, reactivity increases when a request to raise power
decreases the zone levels. Procedures differ in detail from station to station, but
usually these two processes alternate. This avoids the complication of having
simultaneous device operations changing core reactivity.
The following example illustrates the principles common to all procedures. First,
we describe the start up; then we explain why it works and what it achieves. We
will start our description of the process with the reactor significantly subcritical.
Poison removal is stopped. The regulating system is operating and the operator
enters power requests directly into RRS (RRS is in Reactor Leading Mode).
Notes:
CNSC 139
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The operator notes the power level and request a power double the present
level, (that is, 0.3 decades higher). RRS reduces zone levels to increase reactivity.
The operator monitors, and when the zones get down to about the 20% level, he
stops them from draining by asking the regulating system to hold power (at the
present level).
Now poison removal is started again with RRS holding power. As poison
concentration decreases, RRS raises zone levels to compensate, keeping the
reactivity of the core constant. The operator monitors, and stops poison removal
when the zones are high.
These two steps, RRS response to a request to double the present power with
purification stopped, followed by poison removal with RRS holding power, are
repeated until RRS achieves a power doubling when the operator requests it. The
process is then iterated two (or three) more times, achieving three (or four) definite
power doublings in all.
Why does the operator ask for power to double? Why does the operator look for
three doublings of power? Is the reactor critical?
Let us look at this startup process in more detail, starting with the iteration just
before the power doubled the first time. For illustration, we will assume the
reactivity worth of the zone control system over the usual operating range (from
20% to 80% full) is about 4 mk. (This corresponds to 6.67 mk from 0% to 100%
full).
Suppose the operator drives the zone levels up to a little over 65% by poison
removal with RRS holding power and then stops poison removal. Following a
request to double the observed power, RRS lowers the zones to eliminate the
power error. Power increases but as the zones approach 20%, it becomes apparent
that the power is not going to double, so the operator stops the process.
What can we say so far?
First, in dropping from about 65% to 20% the zones insert about 3 mk.
(0.45 × 6.67 mk = 3 mk). Power did not double, so the reactor was initially
subcritical by more than 6 mk.
Now the operator removes poison and drives the zones back up to 60%. There is
no change in reactivity. Suppose that this time when the doubling request is made,
the power doubles when the zones drop to near 20%. The zones have added
almost 40% of their full worth, about +2½ mk. Since power doubled, we conclude
that the reactor was approximately 5 mk subcritical and is now about 2½ mk
subcritical.
Now the operator pulls poison with RRS holding power and drives the zones up to
about 55%, stops poison removal and requests a second power doubling.
Notes:
CNSC 140
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

We know that a power doubling is within the range of the zones and, if the
estimate of 2½ mk subcritical is accurate, the next doubling will occur with
insertion of +1.25 mk. We expect the zones to drop from about 55% to near 35%
(20% of 6.67 mk = 1.3 mk).
Again the operator holds power, removes poison till the zones rise to almost 50%
and asks for the third power doubling. This time the zones drop a little more than
9% to add a little more than 0.6 mk and double power once again. Average zone
level is just over 40% and the reactor is subcritical by about 0.6 mk.
We can see some of the features of this procedure.
• Three doublings allow the operator to fine tune the final zone level,
but that is not the main point of requiring three doublings.
• Three doublings on zone level always brings the reactor close to
critical.
In our example, the reactor is just close enough to critical to satisfy the
requirement for direct regulating system control. If the first doubling in this
example had occurred with a zone level drop of, say, 44% (e.g. from 65% to 21%),
the second doubling would have required a drop of 22%, and the third a drop of
11%. That would not quite satisfy our criterion for saying RRS is in control
(although a 4
th
doubling would certainly be adequate).
One of the nicest features of this procedure is that the reactor cannot actually go
critical! It is a procedure for approaching criticality, cautiously, while ensuring
that the reactor necessarily remains subcritical. As long as the reactor is
subcritical and the operator requests a power doubling, RRS adds only half of the
reactivity required for criticality.
The key to this conservative approach is that when the reactor is nearly critical,
power is increased by adjusting power level, not by direct manipulation of
reactivity devices. The operator cannot initially know exactly how subcritical the
reactor is. Making a positive reactivity insertion into a subcritical reactor by
removing a predetermined amount of poison, or by operating purification for a
calculated time, could cause the reactor to go critical before expected. This
challenges RRS to kick in automatically and take over. If RRS fails, only the
Safety Shutdown Systems can prevent a power excursion.
On the other hand, if you monitor reactor power and look for a power doubling the
reactor does not actually go critical. The increase can be achieved in a subcritical
reactor by adjusting k closer to k = 1.
Restart When the Reactor is Not in the GSS
The reactor is not in the guaranteed shutdown state before restart in two situations.
These are poison override (restart within ~ 30 minutes or so of a trip) and recovery
immediately following a poison out (typically 35 to 40 hours after a trip).
Notes:
CNSC 141
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

In each case, the restart uses the regulating system, but the high, rapidly
changing xenon concentration adds complications (see Section 5.3.4).
Poison Override
Immediate recovery following a trip is possible, if at all, only after an SDS1 trip.
Restart is allowed only after re-poising the safety shutdown system, and for SDS2
this may take nearly as long as the poison-out time.
Assuming restart is an option following an SDS1 trip, the shutoff rods must first
be withdrawn to re-poise SDS1. The reactor stays subcritical because of the
buildup of xenon and because the zones fill and control absorbers drop into the
core on a trip. The control absorbers must also be withdrawn (in banks), followed
by removal of adjusters, one bank at a time, until there is enough reactivity to
overcome the xenon buildup. Suppose criticality is reached (by a decrease in zone
level) following removal of the last bank of adjusters. A request to increase power
begins the xenon burnout process. Zone levels rise again as xenon load decreases,
and the regulating system will request adjuster in-drive (one bank) each time the
average zone level reaches 80%.
Recovery from a Poison Out
In restarting after a poison out, the decay of xenon, which is not under operator
control, determines the rate of reactivity addition. The operating staff must be
ready for criticality, and careful monitoring is required, as predictions of the
poison out time may not be accurate.
During a poison outage, as xenon decays in the reactor, the ion chamber signals
increase until the reactor power reaches its setpoint. At this point, RRS responds
to reactor power measurements. As more xenon decay occurs, the liquid zones
start to fill to maintain the reactivity balance. Once the liquid zones reach their
control limit and the xenon decay continues, reactor power would increase in the
absence of control action. Poison addition and/or adjuster in-drive (if the adjusters
are out of core) are required to maintain the liquid zones in control range.
Once the reactor returns to power, the burnout of xenon is rapid and xenon
simulation by reactor poison is required (see Section 5). This is straightforward
following restart after an SDS2 trip, as the core is heavily poisoned. Poison
removal begins after SDS2 is re-poised, but there will we excess poison in the core
at criticality. Poison burnout approximately matches xenon buildup so purification
can be stopped when power is raised.
One interesting complication is an SDS1 trip immediately following recovery
from a poison out. In this case, there is no xenon buildup, and care is required to
make sure the reactor does not go critical on rod withdrawal.

Notes:
CNSC 142
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

7.5 LOW POWER OPERATION FOLLOWING STARTUP
Usually following criticality, reactor power is ramped into the high power range to
allow electrical power generation. Sometimes, however, extended operation at
low power critical may be necessary. We will discuss the characteristics of this
operating state before going on to high power operation.
As discussed earlier, the critical reactor at a constant low power is necessarily
slightly subcritical because of the photoneutron source. The startup procedure will
have verified that the reactor is within something like 10% zone level of critical,
ready to be made supercritical and ramped to high power as required. The
operator monitors power level, which the regulating system should hold constant,
and zone level, which should change very slowly with changes in the photoneutron
source strength.
The explanation of gradual zone movements requires us to look more carefully at
the photoneutron source.
On return to power, the photoneutrons build back to equilibrium, with the short-
lived ones building up first, and the others following along. Following restart, the
fission products that produce energetic gamma rays that generate photoneutrons
are a mixture of isotopes, some of which are building up from current fissions,
while others remain from previous high power operation.
For example, suppose the reactor went critical at 10
-5
of full power, and power was
then ramped to 10
-3
and held there after a long outage. The source remaining from
previous high power operation is small, having decayed during the outage. Short-
lived fission products emitting energetic gamma rays that produce photoneutrons
are building up, not to the 100% full power level, but perhaps to a level as large as
the source remaining from previous operation. As the source strength increases,
the regulating system will have to reduce k to hold power constant. The reactor
could drift subcritical.
In contrast, the short-lived photoneutron groups do not build to as high a level if
the reactor power stays at 10
-5
of full power. The source will be mainly what
remains from previous high power operation. As this source gradually decreases
(with a 12.8 day half life) the regulating system will gradually reduce zone levels
to keep the reactor at power. With the observed power held constant and the
source decreasing, the subcritical multiplication formula shows that k is drifting
closer to k = 1.
If the power is now ramped to, lets say, 10
-3
of full power and held there for a few
hours, the short lived photoneutron groups will build up to a level comparable to
the left-over source. Now on return to 10
-5
of full power, the source is much
larger, and the subcritical multiplication formula shows that the regulating system
must decrease k below its previous value to hold power.

Notes:
CNSC 143
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

It is quite possible that a reactor that was made critical could, after power
manoeuvres (during which it was clearly supercritical), be subcritical on return to
its initial power level. The liquid zone control system might not have enough
available reactivity to raise power.
A Test for Criticality
Operation with the reactor significantly subcritical is not allowed, except for
transitions to and from the guaranteed shutdown state. The intent is to avoid long
periods of operation in which the regulating system may not be able to effectively
control reactor power. (Long periods of time in the GSS are not a problem as
there are no equipment failures or failure combinations that could make the core
critical.) Startup, the transition between GSS and critical, requires the dedicated
attention of control room staff, following approved procedures.
To help avoid situations in which the reactor drifts subcritical, there is a test to
decide whether a reactor at low power is close enough to the condition k=1 to be
considered “critical”, (that is, can the regulating system control power?). The test
repeats the final steps of taking the reactor critical. The operator requests a power
doubling and if this occurs with only a small change in liquid zone level (for
example, less than a 10% decrease, or a ∆k of about 0.6 mk), the reactor remains
critical. It may be necessary to check this at regular intervals if the reactor is being
operated at a very low power level for a long time.
Raising Power
Section 2 describes and explains power increases following small reactivity
additions at low power where there are no feedback effects. Power initially rises
quickly over less than ½ second (the prompt jump), but only by 1% or so for a
small step reactivity insertion. This is followed by a stable rise with a constant
rate log of power, that is, a constant fractional power increase (percentage of
present power per second = 1/τ).
To raise power, the operator requests a power increase by entering a power
setpoint in the control computer and selecting an appropriate rate of increase. The
formulas in Section 2 model the core response quite well. A reactivity insertion of
0.2 mk or so, achieved by a zone level drop of 3% to 5%, produces a stable rise at
a rate near 0.5% P.P./s (τ ≈ 200 s). This rate, a faster rate than manoeuvres in the
high power range use, will bring power from 10
-4
of full power to between 1% and
10% F.P. in about ten minutes.
As power rises into the range of 1% to 10%, xenon begins to build up and fuel
heating begins to be significant. Xenon buildup is slow; its effect is significant
some hours later. Before power reaches 10% of full power, the temperature
increases of fuel and coolant will have removed a couple of tenths of a mk of
reactivity. Without the active response of the regulating system the power rise
would stop.
Notes:
CNSC 144
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

To continue ramping power, the regulating system will have to continue
decreasing zone level to offset the temperature effects. To maintain a rate near the
original 0.5% P.P./s, net reactivity (the difference between reactivity inserted by
the liquid zones and reactivity removed by increasing temperature) must remain
constant at 0.2 mk or so.
In the power range from below 10% F.P. to 100% F.P., active changes of
reactivity devices are required to change power, or to hold power after a change
(as xenon concentration changes). These reactivity insertions are to offset the
reactivity feedback effects. The regulating system never needs more than about
0.1 mk net reactivity insertion to raise power at an acceptable rate in the high
power region.
The power coefficient results mainly from the fuel temperature coefficient. Its
effect is approximately linear, typically about -0.3 mk for each 10 % of full power
increase in power.
The equilibrium xenon load builds slowly to a relatively low equilibrium
concentration below a few percent of full power, increasing strongly in the range
of 10% to 60% full power. Above this range, equilibrium xenon is always within
about a mk of its full power equilibrium value (see Section 5).
Notes:
CNSC 145
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

REACTOR OPERATIONS AT HIGH POWER
8.0 INTRODUCTION
For operations producing the rated power output, it is important to limit the power
from individual channels or bundles to safe levels. We examine four different
ways of flattening the overall flux shape: addition of a reflector, use of adjuster
rods, bi-directional refuelling, and differential burnup.
We also describe the role of the liquid zone control system in maintaining a zone-
to-zone flux balance, thus limiting high zone flux that could result from flux tilts
and oscillations. In addition, we review some practical aspects of fuel
management that limit local peaks (“hot spots”). Reactivity device configurations
also affect local peaks, so we will look at the reactivity effects of rod drives.
The Regional Overpower System (ROP) protects the reactor from high bulk power
and high local power. We will introduce the principles that underlie this system.
One of our main concerns in operating a power reactor is to ensure that fuel
bundles are not over-rated, leading to fuel damage. Since the fission rate, and
hence power generation, in a bundle is proportional to the thermal neutron flux at
the bundle position, we should have some idea of how the flux varies from one
part of the reactor to another.
Solving the coupled equations described in Section 1 for a finite, homogeneous
reactor (one with a uniform mixture of fuel and moderator) leads to smooth flux
shapes gently varying from a central peak to zero at the boundary. The CANDU
reactor is a heterogeneous system, where fuel is lumped into bundles to reduce the
U-238 resonance absorption. Because thermal neutrons are strongly absorbed in
concentrated fuel regions, the thermal flux is markedly depressed in the fuel itself
relative to the moderator.
The overall flux shape in the CANDU core is generally high in the middle and
falls off to the edges, but techniques to be described here extend the high flux over
a much larger region of the core, while suppressing the central peak. There are
local “peaks and valleys” superimposed on this overall shape.
8.1 FLUX FLATTENING
It is possible to calculate the ratio of the average thermal flux in the core to the
maximum thermal flux for the homogeneous core, using the calculated flux
variation in a cylindrical reactor. The ratio turns out to be
275 . 0
MAX
AV
=
φ
φ

Notes:
CNSC 146
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This low value for the average to maximum flux raises problems. The
total power output of the reactor is proportional to the average flux, so it is
advantageous that this be as high as possible. However, a maximum heat rating
(to avoid risking damage to the fuel) sets an upper limit on the flux. This limit
applies to fuel at the core centre of a homogeneous reactor, where the flux has its
maximum value. With φ
MAX
set by safety considerations and φ
AV
equal to only
27.5% of φ
MAX
, the fuel away from the central peak contributes far less than its
potential share of the power.
The solution is to increase the ratio of φ
AV
to φ
MAX
. Flux flattening is the name
given to the improvement of this ratio. In the remainder of this section, we will
discuss four ways of flattening the thermal neutron flux in the radial and axial
directions. These are:
• addition of a reflector (for radial flattening);
• adjuster rods (for axial and radial flattening);
• bi-directional refuelling (for axial flattening);
• differential burnup (for radial flattening).
The reactor design incorporates the first two of these; fuelling achieves the last
two. The net result is an increase of the average flux from 27.5% to about 55% or
so of the maximum, approximately doubling the heat output of the reactor without
increasing any channel or bundle power above the peak allowed value.
The Reflector
Each CANDU reactor has a radial reflector that consists of a radial extension of
the moderator that provides a layer of about 70 cm of heavy water. This layer of
heavy water surrounds the core and reflects many leakage neutrons back into the
core. This produces two effects: it reduces leakage, and increases the flux near the
core boundary (flux flattening).
Figure 8.1 illustrates the effect of the reflector in flattening the thermal flux in the
radial direction. The theoretical flux shape for a bare core is labelled “flux without
reflector”. The lower flux shape with the reflector in place is normalized to the
same total power output as the bare core flux. The higher flux shape with the
reflector in place shows the power raised so that the maximum flux in the core is
the same as it was for the bare core.




Notes:
CNSC 147
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Fast neutrons that are thermalized in the reflector cause the humps there.
They “pile up” in the reflector because they are less likely to be absorbed there
than in the core.
Adjusted
flux with
reflector
flux without
reflector
Flux with
reflector
(at same power)
Plot
-300 -200 -100 0 100 200 300
Reflector Reflector
Distance along
radial direction

Figure 8.1
Effect on Reflector on Shape of Radial Flux
Adjuster Rods
Adjusters are rods of a neutron-absorbing material normally inserted into the
central region of the reactor. They have the dual function of flattening flux (both
axially and radially) and of providing excess reactivity when called upon. The
name “adjusters” is comes from the flux flattening function—adjusting the flux
shape.
Notes:
CNSC 148
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003


Flux with
Adjusters
Inserted
Flux with
Adjusters
Withdrawn
Adjusters
Positions
A B
END TOP
R
e
l
a
t
i
v
e

F
l
u
x
Distance Along Chord AB (cm)
100200 300 400 500 600 700 800
0
0.2
0.4
0.6
0.8
1.0
1.2

Figure 8.2
Flux Flattening Produced by Adjuster Rods
Figure 8.2 illustrates flux flattening in the radial direction by the adjuster rods.
The figure also shows the flux shape with adjusters withdrawn. The curves for the
flux with and without adjusters show the same peak flux, since this is what
imposes the limit to avoid fuel damage.
As the regulating system withdraws the adjusters from the core, it simultaneously
increases liquid zone levels to keep the power output constant. This would
preserve the areas under the two curves. To keep the peak flux from increasing,
the operator would need to reduce the reactor power setpoint before the rods come
out of core. This reduces the height of the curve labelled “flux with adjusters
inserted”. Adjuster withdrawal then produces the more peaked shape shown, with
the peak height matching the original peak height. It is clear that a reactor with
adjusters inserted produces higher power for the same maximum flux.
Bi-Directional Refuelling
Fuelling a channel involves removing high burnup bundles from a fuel channel
(usually four or eight bundles) and replacing them with fresh fuel. Partly burnt-up
fuel then occupies one end of the channel, with fresh fuel in the other end. The
fission rate in the new fuel is higher so it produces higher thermal flux in that end
of the core. If all the refuelling were done from the same end of the reactor, this
would eventually create a marked asymmetry in the flux with a high peak near the
end with fresh fuel. To avoid this, we refuel adjacent channels in opposite
directions, thereby keeping the flux shape essentially symmetrical.
Figure 8.3 shows that the addition of the asymmetric flux shapes produced by
channels refuelled in opposite directions produces an axial flux shape that is
somewhat flattened. The curve labelled “theoretical shape” in this figure assumes
replacement of all fuel bundles during a refuelling operation.
Notes:
CNSC 149
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

This does produce a symmetrical shape, but to get the same total power
output, the peak flux would need to be much higher, as shown.

1 2 3 4 5 6 7 8 9 10 11 12
Total Flux
Theoretical Flux
Shape with Uniform Fuelling
Flux from
Channels Fuelled
Right to Left
Flux from
Channels Fuelled
Left to Right
Bundle positions along
channels

Figure 8.3
Effect of Bi-Directional Refuelling in Flattening Axial Flux Shape
The flattening effect is greater with a four (or two)-out-of-twelve refuelling
scheme, but even the eight (or ten)-out-of-twelve replacement produces some flux
flattening.
Differential Burnup
Differential burnup is a method of radial flux flattening produced by fuelling
strategy. The fuelling engineer allows fuel in the inner zone of Figure 8.4 to reach
higher burnup than fuel in the outer zone. To keep the reactor critical, fresher fuel
in the outer core supports the high burnup fuel in the inner core. With less fissile
material and more fission product absorption in the inner zone, the fission rate
distribution is less peaked, so the flux shape is flatter, as shown in Figure 8.4.
Notice that the actual dwell time of fuel in the inner zone is less than in the outer
zone. Although the burnup is higher, high burnup occurs more quickly in the
higher flux region.
Notes:
CNSC 150
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Reactor Face
Inner Zone
Outer Zone
Flux with
Differential Fuelling
Flux without
Differential
Fuelling
Inner
Zone
Radial Position

Figure 8 4
Flux Flattening Produced by Differential Fuelling
The BNGSA reactors do not have adjusters, which results in some loss of
operating flexibility. However, by eliminating the burnup loss caused by adjusters
these reactors achieve high fuel burnup. Differential fuelling flattens flux in the
radial direction and two bundle shifts on refuelling in the inner zone helps flatten
flux axially. The higher parasitic absorption in the inner core has much the same
effect as adjuster rods in the inner core.
8.2 FLUX SHAPE DETAILS
The actual flux shape in the reactor core differs somewhat from the overall smooth
flux shape just described. Neutron absorption in fuel and in reactivity devices
distorts the smooth flux shape. For an equilibrium fuelled CANDU, the overall
flux shape is nearly constant during operation, but there are local “bumps” and
“hollows” that change strength and location because of changes in xenon
concentration, because of fuel burnup and on-power fuelling, and because of
changes in the configuration of the reactivity devices.
Local flux peaks must be controlled so that safe operating limits on fuel bundle
power and channel power are not exceeded. Peaks can affect fuel power directly
or by distorting measurements of power. In routine operation, a good fuelling
strategy coupled with effective power regulation limits the peaks.
Notes:
CNSC 151
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Anything that could produce an unanticipated local peak, such as a stuck
rod, or an uncontrolled xenon oscillation requires attention.
Flux Depression around Absorbers
Figure 8.5 shows the thermal neutron flux depression across a CANDU fuel
bundle. Notice that the reduced flux extends beyond the boundary of the bundle.
The thermal flux just outside the bundle results from the random motions of
neutrons moving into the region from all directions.


Figure 8.5
Depression of the Thermal Flux in a Fuel Bundle
The presence of a nearby absorber reduces the source of neutrons from that
particular direction, so the flux is reduced. This effect is seen for any absorbing
material in the core. Not only is the flux reduced in the region where neutrons are
absorbed, it is reduced in the surrounding region as well. The flux depression gets
less the further you are from the absorber.
Zone to Zone Flux Flattening
The calibrated zone power signals represent the average power output from each
zone. The regulating system raises or lowers individual zone levels to keep the
average flux in each of the fourteen zones close to the overall average.
The regulating system phases in spatial correction between 15% and 25% zone
power, so full spatial control is available when power reaches the threshold for
onset of xenon oscillations, about 25% full power. Xenon disturbances below this
power do not cause self-sustaining oscillations.
We can estimate the flux level below which xenon disturbances are not a problem.
One of the requirements for an oscillation (or transient xenon effect) is a xenon
burnout rate significantly greater than the xenon decay rate.


Notes:
CNSC 152
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The flux that gives equal xenon burnout and decay rates is
Xe Xe Xe
Xe
a
N N λ = φ σ so
1 2 12
2 18
1 5
Xe
a
Xe
s cm 10 6
cm 10 5 . 3
s 10 12 . 2
− −

− −
× =
×
×
=
σ
λ
= φ
CANDU average fuel flux at full power ranges near 7 to 9 × 10
13
cm
-2
s
-1
, so this is
just under 10% of full power.
A local, high power xenon effect not accounted for in Section 5 is the effect of
fuelling a high burnup channel. The fresh fuel (with high fissile content and no
fission products) causes an immediate increase in flux locally. The high local flux
increases burnout in neighbouring channels. Over the next several hours, while
xenon is building into the fresh fuel, the high flux continues to burn out xenon in
nearby channels. This initially produces a local hot spot. The liquid zones, in
responding to an average flux increase in the zone, have some effect in limiting the
size of the peak. The hot spot is self-limiting over a number of hours. The
buildup of iodine in the neighbouring channels results in increased xenon
production and neutron absorption, and so does the buildup of xenon in the fresh
fuel.
Usually these fuelling transients cause no particular problem, as long as the
problem is not compounded by inappropriate additional fuelling or device
movements. At high power, such problems would be detected promptly, and
technical advice solicited to correct the problem before additional fuelling takes
place, making things worse.
Reactor fuelling should be done with the reactor at high power, so that corrective
action can be taken if there are unforeseen reactivity effects such as a hot channel,
or flux tilt. These abnormalities cannot be detected with the reactor shut down. A
latent flux distortion due to shutdown fuelling activity could go undetected.
Shutdown fuelling can be done on a case by case basis with management approval,
but when the reactor is returned to power careful monitoring for flux distortions is
required as power is raised.
Local Power Peaks
The liquid zone control system, designed and operated to regulate bulk power, and
to adjust zone average power, cannot effectively eliminate a local “hot spot” that
might develop during operation. Such detailed flux shaping is accomplished
routinely by selecting appropriate channels for on-power fuelling. The next
subsection discusses several considerations taken into account in selecting
channels for fuelling, including limiting local power peaks.
Notes:
CNSC 153
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

To describe the local peaks and valleys in the flux distribution we compare
the actual flux distribution with an idealized reference flux shape. The reference
flux shape derives from the shape that would exist if day-to-day variations could
be averaged out. (This is variously called the time-averaged flux shape or the
continuous fuelling model flux shape.) There is no possibility of achieving this
theoretical shape, as burnup and fuelling in normal operation produce local
fluctuations about the average. The fuelling engineer consults a channel power
map based on the reference shape as an ideal “target” for fuelling.
Deviations in flux from the reference shape are called fuelling ripple. Fuelling
ripple, defined for each channel, is the ratio of the actual (measured) channel
power to the reference channel power. (The off-line fuelling code generates a map
of reactor fuel channels with ripple values for each.)
For example, if a particular channel has a reference power of 6.5 MW (meaning
that over many years of routine, full power operation the analysts expect power
from this channel will average 6.5 MW), and the actual power is 6.2 MW before
refuelling, the ripple is 6.2/6.5 = 0.95 for that channel. If the channel power
reaches a steady 6.9 MW after refuelling, the ripple is then 6.9/6.5 = 1.06.
Fuelling ripple changes continually with core conditions.
Given the fuelling ripple for every channel in the core, there will be one highest
value of the ripple. This value is the channel power peaking factor (CPPF).
(Stations may discount high ripple in the outer one or two rows of channels at the
edge of the core when selecting the CPPF channel. Stations may increase the
“measured” CPPF to include uncertainties in measurement and analysis.)
All channels and bundles in the core operate below their safe operating limit:
• if bulk power is controlled,
• if the overall flux shape is kept adequately flat, and
• if fuelling keeps local peaks acceptably low.
The regulating system accomplishes the first two of these, provided the operating
staff ensures that zone levels operate within their normal operating range, and
there are no flux distortions from unusual reactivity device configurations. The
safety shutdown systems back up normal regulation. A later section discusses the
Regional Overpower Protection System (ROP) - sometimes called the Neutron
Overpower Protection System (NOP).
The third item in the list depends on selecting and fuelling the right channels.
(Fuelling also affects the second item on the list). The next subsection describes
how the fuelling engineer selects channels for fuelling.
Notes:
CNSC 154
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Effects of Fuelling
On-power fuelling is used for long-term reactivity management and for flux
shaping in CANDU reactors. The Fuel Engineer at the station is responsible for
ensuring that the optimum fuel cycle is used. A computer program that simulates
the core helps select channels for fuelling. Among its many calculations, the
simulation determines the axial and radial power distributions, the burnup of each
bundle, and the excess reactivity.
A comparison of power output predictions with accurate thermal power
measurements (Heat Balance) checks the validity of these calculations. If there
are discrepancies, the program is “tuned” by appropriate adjustments to the
physics data to obtain satisfactory agreement between the computer code and the
measured quantities.
The fuelling engineer reviews the output of the fuelling program and then selects
channels for fuelling based on
• safe operation,
• reactivity requirements,
• flux shape control,
• fuel cost, and efficient use of the fuelling equipment.
Regulation and protection of the core in day-to-day normal operation depend
crucially on the appropriate choice of channels for fuelling.
On-power fuelling principles and their explanations follow:
• Do not operate with defected fuel elements in the core.
Fuel defects release radioactive fission products to the coolant. This increases
chronic dose to station staff. Radioactive releases increase on power ramps. A
power change with just a few defective elements in core could result in I-131 in
the coolant exceeding allowed limits—limits based on possible accidental release
to the public. Defective fuel elements in core also increase the radioactive fission
product release from the fuel following a major accident. In addition to these
safety issues, contamination of the fuel defect detection and location systems
makes them less sensitive to subsequent fuel failures.
• Regularly replace enough fuel to keep the average zone level near
midrange.
In a CANDU operated at high power without fuelling, the liquid zones will
typically drop about 5% each day to offset the loss of reactivity from burnup.
High or low average zone level can make bulk power regulation difficult.

Notes:
CNSC 155
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

There may not be enough available range to compensate for the xenon and
temperature reactivity effects that occur during and following a power manoeuvre.
• Locate the new fuel so that the overall flux shape stays flat and so
that individual zone levels do not have to deviate too much from
the average to keep it flat.
Differential fuelling and fuelling adjacent channels in the opposite direction assist
in keeping the overall flux shape flat, as described earlier in this section. In
addition, fuel is added preferentially to zones with low zone level to raise them
toward the average.
At times, it is not possible to fuel optimally because of technical problems. The
liquid zone control system is able to keep axial flux shape flat after fuelling
several channels consecutively in the same direction. Similarly, after fuelling
several channels consecutively in the same zone, zone to zone flux is equalized by
liquid zone level adjustments. In either case, however, individual zone levels will
deviate significantly from the average. Very high or very low individual zone
levels must not be allowed or spatial regulation is lost. This could lead to a
channel or bundle power above the license limits, or an NOP trip or a setback
could occur.
• Distribute the fuel to minimize fuelling ripple.
Fuelling near a low burnup channel can result in both channels building to near
maximum reactivity (the plutonium peak) at the same time. The computer
simulation helps the fuelling engineer anticipate and avoid hot spots, which could
necessitate power derating to maintain safe operation. The CPPF could increase,
causing reduced margin to trip or, in an extreme case, inadequate trip coverage.
Channel or bundle power could increase to the license limit in the hot region.
• Replace fuel with the highest burnup.
It makes economic sense to fuel channels so that the highest burnup fuel is
discharged. Typically, this also gives the highest reactivity gain per channel
fuelled.
The reactivity gain from fuelling is highest in the central core and falls off to the
edges. Typical fuelling strategies insert more bundles per visit to outer core
channels than to inner core channels. This results in fewer fuelling machine visits
to the outer core, even though the average discharge burnup is lower in the outer
core than in the inner core. More frequent visits to the inner core channels may be
justified by the smaller fuelling ripple that results from inserting only a few fresh
bundles.
Replacing high burnup fuel reduces fuel cost; high reactivity gain per channel
fuelled reduces frequency of fuelling operations and thus decreases fuelling cost.
Notes:
CNSC 156
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

• High burnup bundles should be distributed in the core so
they are not exposed to high flux and large power ramps.
Old bundles at relatively high power are at risk of failure by stress corrosion
cracking if stressed by a large power ramp. Such a ramp could be caused by
reactivity device movement or by fuelling. Good overall fuelling strategy,
together with correct sequencing of adjuster bank withdrawal and imposition of
bulk power limits on adjusted flux shapes have contributed to good CANDU fuel
performance. CANDU 37-element fuel bundles have shown no vulnerability to
ramp defects, and 28-element fuel bundles only occasionally fail in this way.
Regular fuelling is required to maintain core reactivity. The above principles
guide the fuelling strategy, but may at times contradict one another. For example,
fuelling to replace the highest burnup fuel may increase ripple; outer core visits are
required for overall flux flattening, but do not give as large a reactivity gain as
fuelling the inner core. The following is a typical optimum fuelling strategy, in
the likely order of priority:
• Remove fuel with known defects at the earliest opportunity.
• Fuel so that the fuelling ripple is controlled
- Keep the CPPF at an acceptably low value.
- Ensure adequate operating margin to license limits and trips
• Fuel evenly across the core:
- Visit the inner and outer core with the right frequency.
- Alternate the fuelling direction.
- Avoid fuelling consecutively into the same zone.
• Fuel channels with high discharge burnup and high reactivity gain.
All fuelling operations are documented so the corresponding changes in the reactor
core are accounted for in the fuelling simulation program, and so all bundle
locations can be traced.
Reactivity Device Movements
A control rod or shutoff rod is typically a cylindrical rod made of some material
with a high absorption cross-section for thermal neutrons, for example, cadmium
in stainless steel. The worth of a rod is the reactivity change produced when the
rod is inserted in the reactor. This depends most obviously on the absorbing
material in the rod. It also depends significantly on the position of the rod in the
core. The same rod will have a different reactivity worth in different locations.
From time to time individual shutoff rods may become stuck in a reactor at high
power, for example on a routine partial rod drop test to verify fast insertion times.
Notes:
CNSC 157
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Occasionally, high zone levels cause a bank of control absorbers to drive
into the core to reduce zone levels.
Let us consider the worth of a single shutoff rod inserted into the high-flux region
at the centre of the core (Figure 8.6). Since the rod is an effective absorber of
neutrons, the thermal flux in the region around it goes down significantly. If the
reactor continues operating at the same bulk power with the rod inserted, the
regulating system must compensate for the decreased flux in the central region by
raising the flux in the outer regions, as shown in Figure 8.7. The result is to
displace the flux towards the edge of the core, which will lead to a greater thermal
neutron leakage.
Core

Figure 8.6
Control Road Inserted into Central Core Region
Flux
Position on Axis
Flux
Position on Axis
Before After

Figure 8.7
Thermal Flux Shape Before and After Insertion of a Rod
Notes:
CNSC 158
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

Core
Flux
Position on Axis

Figure 8.8
Thermal Flux Shape when Rod is Inserted Near the Edge of the Core

The worth of a rod therefore results from a combination of two effects:
• increased neutron absorption in the region where it has been
inserted, and
• increased leakage caused by the flux distortion produced by the rod.
The size of each effect depends on the placement of the rod in the core. Suppose
that the rod insertion is near the edge of the core as shown in Figure 8.8. The
absorption effect is obviously less than in the previous case because there are
fewer neutrons to be absorbed in this lower flux region. The regulating system
therefore increases the flux in the rest of the core by a relatively smaller amount to
maintain bulk power. The flux distortion and the increase in neutron leakage will
also be less than for the rod insertion into a high-flux region. The overall effect is
that the rod worth will be considerably less at the edge of the core than in the
central region.
Differential Rod Worth
Let us now examine how the reactivity effect of a shutoff rod or control absorber
rod varies as it is gradually inserted into the reactor. We can describe this
quantitatively by introducing the differential rod worth, defined as the change in
reactivity worth with, say, each extra millimetre of rod added to the core. The
total reactivity worth is the worth of the fully inserted rod.
Figure 8.9 shows the variation of the reactivity worth of the rod as a function of its
position. The total reactivity of this rod, fully inserted, is about 2 mk. Starting
with the rod completely withdrawn, the initial movement produces relatively little
effect, since the rod is entering a region where the neutron flux is low. As it enters
further into the core, the differential worth (worth per millimetre) steadily rises,
reaching a maximum as the leading end reaches the core centre. The slope of the
Figure 8.9 curve is the differential worth, which, as you can see, is greatest when
the end of the rod reaches the centre line of the core. As the end of the rod
Notes:
CNSC 159
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

progresses beyond the centre, the differential worth decreases again,
reaching a minimum at the fully inserted position.
400 350 300 250 200 150 100 50 0 -50 -100 -150 -200
2
1
0
Distance relative to centre line (cm)
R
o
d

w
o
r
t
h

(
m
k
)

Figure 8.9
Reactivity Worth of Rod as a Function of Position
In other words, the rod is most effective when it reaches the high flux central
region of the core. This helps explain the need for the spring assisted start that
accelerates the shutoff rods when they drop. It also accounts for the great
importance placed on testing the shutoff rods to verify the rates of insertion.
In passing, we might note that the partial movement of a rod or a bank of rods into
the core would produce the same distortion of the neutron flux shape as described
in the previous section, but in the vertical direction. The flux in the upper part of
the core will be reduced relative to the flux in the lower part.
Note that the shutoff rods and control absorbers are seldom in the core at high
power; they usually drop into core to reduce power rapidly to a low level. The
regulating system does not routinely drive control absorbers for reactor regulation.
The operator can easily avoid the flux distorting effect of the control absorbers by
using the poison addition system to reduce zone levels before rod drives occur.
On the rare occasion when the control absorbers drive into core to assist the liquid
zone system, the zones drain as a bank of two control absorbers drives in. The
way the “bite” of the rod changes with depth has little operating significance; it
simply produces a varying rate of zone level decrease.
Rod Shadowing and Anti-Shadowing
The worth of a rod placed in some location depends on whether other rods have
already distorted the overall flux shape. To illustrate, suppose that we have
inserted a single rod into a high-flux region of the core, as shown in the top portion
of Figure 8.10. The regulating system, holding power at setpoint, compensates for
the insertion of the rod, distorting the flux shape as shown.
Suppose that we now insert a second identical rod quite close to the first one, as in
the middle illustration in Figure 8.10. Since it will be entering a region where the
first rod has depressed the flux, it will absorb fewer neutrons than it could have if
Notes:
CNSC 160
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

the other rod were not there. In addition, the presence of the second rod
reduces the flux in the neighbourhood of the first one, thereby reducing its worth
compared to the situation where it was the only rod in the core.
Thus, if each rod on its own had a worth of, say, x mk, the worth of the two in
combination will be less than 2x. The reduction in individual worth of each rod
because another is nearby is known as rod shadowing.
Consider now what would have happened if the second rod had been inserted into
a region far away from the first one, for example, at the peak of the distorted flux
distribution produced by the former, as in the bottom illustration in Figure 8.10.
Because it is going into a region where the first rod and the regulating system
actions have increased the neutron flux, it absorbs more neutrons than if it were
the only rod in the core. Again, the effect is symmetrical and the worth of the first
rod increases. The combined worth of the two rods is then greater than 2x. This
increase in the worth of each rod due to the presence of the other is known as rod
anti-shadowing.
Core
Flux
Position on Axis
Core
Flux
Position on Axis
Core
Flux
Position on Axis

Figure 8.10
Rod Shadowing and Anti-Shadowing
Notes:
CNSC 161
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

In a CANDU, the effects of rods on one another are most noticeable for
the adjuster rods. These are typically arranged in six or seven symmetrical banks
of from two to four rods each, with a nominal reactivity worth of a little over 2 mk
per bank. The adjusters are made of relatively mild absorbing materials, so they
are less flux distorting than the shutoff rods and control absorber rods.
Adjuster rods are normally removed from the core (and returned) one bank at a
time in an approved sequence. The approved sequence takes account of rod
shadowing and anti-shadowing effects so that flux distortions caused by rod
interactions do not produce unacceptable flux peaks.
Removal of banks of adjusters in the approved sequence makes the overall flux
more peaked—one reason for having the rods in-core is flux flattening. There is
an upper limit on bulk reactor power for each allowed configuration of adjuster
rods, to limit maximum bundle and channel power.

Equipment problems may on occasion make it impossible to drive the rods in the
correct sequence, or may prevent all the rods in a bank from driving. Analysis in
advance is required to determine the correct bulk power limit and to make sure
local flux peaks are acceptable.
Flux peaking is more severe when adjusters are used as a reactivity shim (for
example, if fuelling is not available) than when used for xenon override (for
example, following a power decrease). The maximum increase in xenon during
the transient is in the highest flux regions of the core, where iodine inventory was
highest. This is where the adjusters are located to flatten flux. The increased
xenon absorption in these high flux regions helps flatten flux in the region.
Operating manuals for some stations may have different bulk power limits on
adjuster removal, depending on how the rods are used.
When a bank of adjusters begins to drive, it continues moving until fully out (or
until fully in, if it was originally out of core). The regulating system requests
adjusters to drive out when there is not enough excess reactivity, that is, when
zone levels are low. (RRS requests in-drive on high zone level, but before the
level is high enough to call for control absorber in-drive.) The immediate effect of
a bank of rods driving out of core is an offsetting zone level increase.
8.3 HIGH POWER PROTECTION
We will consider high power protection under two main headings:
• Regional Overpower Protection (ROP)
• Licence limits on bundle and channel power
Power Rise Limited by NOP/ROP
Notes:
CNSC 162
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

The ROP systems (one for each shutdown system) provide primary or
backup trip coverage for a variety of process failures that cause excessively high
power. The system caters particularly to power increases that are too slow to
produce a rate log trip.
Each ROP system includes three arrays of self-powered in-core detectors to
provide three trip channels. For adequate protection, at least one detector in each
channel should see any event that produces high bulk power or high local power.
Specifically, the trip setpoints ensure that a reactor trip occurs before heat
production causes either centerline melting or fuel dryout anywhere in the core.
Centerline melting occurs if the rate of heat production in a fuel element exceeds
the capacity of the pellet to transfer heat to the fuel sheath and into the coolant, so
that the fuel centerline temperature reaches the melting temperature for UO
2

(about 2750°C). Fuel dryout occurs when the fuel sheath temperature gets so high
that excessive boiling of coolant at the fuel surface produces a vapour barrier that
limits heat removal from the fuel element.
Dryout results in rapid fuel heating followed by centerline melting. Typically,
centerline melting occurs if a bundle power is too high (because of high local
flux), and dryout occurs if the channel power is too high (because of high total
power output along the channel). Analysis demonstrates protection from both.
The system designers cannot know the initial power in each channel, as these vary
from day-to-day because of fuel burnup and refuelling. The designer also does not
know what configuration the reactor will be in when a specific event occurs that
requires a trip. Finally, the designer must provide coverage for a wide range of
initiating events. In the analysis, a trip occurs when any channel power, initially at
the reference channel power reaches dryout power (or any bundle reaches
centerline melting). Protection for a power rise in a real core (that is, in a reactor
with fuelling ripple) occurs because the ROP detector signals are raised (up
calibrated) to protect the channel with the highest ripple.
To explain how the trip setpoints are determined and how trip coverage is effected,
three items are discussed.
• The design basis set of flux shapes
• Determining the required trips for the reference flux shape
• Taking account of fuelling ripple (CPPF)
The Design Basis Set of Flux Shapes
Trip coverage analysis considers a very large number of events from a variety of
initial reactor configurations, at minimum several hundred scenarios. These are
events like loss of bulk power control, producing a slow uniform increase in power
across the core, or draining of a single zone, causing a bulk power increase with a
high local peak superimposed, etc. These and many other upsets are combined
Notes:
CNSC 163
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

with various initial distortions of the reference flux shape core conditions
(for example, reference core at steady high power, reference core with a xenon
transient in progress, reference core with a rod stuck in core, etc.) The ROP
system must provide overpower protection for a large number of initial flux shapes
that could arise due to normal or abnormal movement of reactivity devices,
together with changes in xenon concentration. The design basis set of flux shapes
is a comprehensive set of flux shapes representing all analyzed upsets.
Reference Core Trip Coverage
Each detector must have a trip setpoint such that at least one detector in each array
trips before any fuel channel reaches dryout, or before onset of centerline melting,
whichever comes first. Conceptually this is quite simple. A computer simulation
begins with the reactor in the reference flux shape and simulates each of the
analyzed upsets in turn. It allows each design-basis flux shape to increase until the
highest channel power reaches the power level corresponding to onset of dryout
(or the highest bundle power corresponds to the centerline melting temperature).
The computer then notes the thermal flux level at all the in-core detector locations.
As the computer analyzes each of the design basis flux shapes it sorts through the
detector readings and selects the least restrictive set of trip setpoints (that is, with
maximum margin to trip) that provides the necessary trip coverage.
As more and more restrictive accidents are analyzed, trip setpoints for all detectors
move downward, perhaps into the range of 117% to 119%. In the early stages of
the design, adjustments to detector locations and their distribution over the arrays
improve trip coverage while optimizing the operating margin (margin to trip). The
analysts may even make small changes to the reference flux shape (making the
reference flux shape slightly different from the time-averaged flux shape). At the
end of the process, each ROP system has a set of trip setpoints that provides
complete coverage for all incidents in the design basis set, provided the reactor
fuelling has produced the reference flux shape.
Effect of Fuelling Ripple
The ROP/NOP trip setpoints protect the idealized reference core, as just described.
In the real core, fuelling ripple may result in the vulnerable channel (for a
particular upset) with a higher power than the reference. The detector expected to
provide trip coverage could be reading below the reference value. As power
increases uniformly, the vulnerable channel reaches dryout (or a bundle reaches
centerline melting) before the detector reaches its trip setpoint. The worst case
situation would be if the vulnerable channel was the CPPF channel (the CPPF
channel has the highest ripple in the core). This difficulty is solved by recognizing
that the actual detector readings don’t matter, only the margin to trip matters. The
safety system ROP detectors do not need to read 100% with the reactor at full
power. Instead, all detectors are adjusted so that they read 100% × CPPF (if the
reactor is at 100% of full power). Setting the reading to 100% takes care of the
Notes:
CNSC 164
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003

detectors reading low because of fuelling ripple. Increasing this by the
CPPF preserves the margin to trip for the worst case scenario. Trips will occur
early if the vulnerable channel is not the CPPF channel.
Poor fuelling can produce a high CPPF that results in a small margin to trip.
When this happens, the NOP trips can seem very restrictive, with power well
below any possibility of dryout or centerline melting. Keep in mind that for some
particular slow loss of regulation accident there is a channel that could reach
dryout just as the trip setpoint is reached.
If the channel ripple is higher than the CPPF because of some local flux distortion
resulting from an unexpected device configuration (perhaps compounded by xenon
effects) due, for example, to a shutoff rod stuck partly inserted into the core, trip
coverage may be inadequate. The design basis set may include some of these
highly unlikely device configurations that give extreme flux shapes. If these
configurations require trip settings that are too restrictive in normal operation, they
are protected by a hand-switch adjustment that reduces the trip setpoints. (The
operator would need to recognize that he is in an abnormal configuration, reduce
power appropriately, and adjust the hand-switch).
Inevitably, many possible reactivity device configurations are not in the design
basis set. Another hand-switch setting that reduces the trip setpoints even further
protects the associated flux shapes, the unanalyzed flux shapes. The analysts
check that this trip setpoint provides coverage for a small bounding set of extreme
flux shapes, not associated with any particular upset.
We end this section with one final comment about trip coverage. CANDU
reactors with 37-element fuel reach dryout conditions before any bundle in the
core reaches centerline-melting temperature. The thinner fuel elements (compared
to 28-element fuel) provide more total surface area for cooling, and less thickness
of uranium dioxide (UO
2
is a poor thermal conductor). CANDU reactors with 28-
element fuel are much more likely to reach centerline-melting temperatures before
any channel reaches dryout power. The 37-element bundle reactors are described
as channel power limited, the 28-element bundle reactors as bundle power limited.
License Limits
An important supplement to the information in the previous section is that the
ROP/NOP systems do not prevent the violation of license limits on bundle and
channel power. For example, a typical license limit on channel power for
37-element fuel is 7.1 MW. The channel power that will produce dryout,
assuming normal coolant conditions, is almost 30% higher than this. Obviously,
in this example, a trip intended to prevent dryout may not prevent power
increasing beyond the license limit.
The ROP/NOP trip ensures the integrity of the fuel channel by reducing power
before there is risk of fuel disassembly that could pose a risk to the channel. An
Notes:
CNSC 165
Science and Reactor Fundamentals – Reactor Physics
Technical Training Group

Revision 1 – January 2003
intact heat transport system guarantees that fission products released to the
coolant will not reach the public. The bundle and channel power license limits arise
from quite different considerations. Safety analysis demonstrates that radiation
releases to the public is within stated limits, and shows adequate primary and backup
trip coverage for a whole variety of different upsets. The analysis depends
sensitively on the fuel power before the analyzed upset. Adequate trip coverage and
protection of the public have not been demonstrated for operation with fuel power
that exceeds the values assumed in the analysis; these values establish the license
limits.
Operation with bundle or channel power above the license limits is operation with
excessively high power. Enforcement of the license limits requires the joint efforts
of the fuelling engineer, the fuelling crew, and the authorized staff. The fuelling
engineer routinely checks the slow evolution of steady state bundle and channel
powers with the off-line fuelling code. The code helps select channels for fuelling
that provide an adequate margin to the license limits in normal operation.
How does the authorized staff know that a license limit has been violated? There is
no direct measurement or panel indication, so the operator does not “know” unless
told by the fuelling engineer. However, if the bulk power output is below the license
limit and the flux is “flat enough”, and the fuelling engineer and fuelling crew have
put bundles into the correct locations, then the bundle and channel powers will be
within their limits.
This means routine monitoring of device positions and zone levels is part of the
enforcement process for license limits on bundle and channel power. Conversely, if
the flux shape is not standard, (an unusual tilt—e.g. with zones limiting, or rods
stuck in core or some combination of rods giving an off normal configuration) then
the operator has little choice but to assume that channel power limits (or bundle
power limits) may be exceeded, unless analysis of the particular configuration
demonstrates otherwise.


Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.